CN109829920B

CN109829920B - Image processing method and device, electronic equipment and storage medium

Info

Publication number: CN109829920B
Application number: CN201910138465.6A
Authority: CN
Inventors: 高云河; 黄锐
Original assignee: Shanghai Sensetime Intelligent Technology Co Ltd
Current assignee: Shanghai Sensetime Intelligent Technology Co Ltd
Priority date: 2019-02-25
Filing date: 2019-02-25
Publication date: 2021-06-15
Anticipated expiration: 2039-02-25
Also published as: CN109829920A

Abstract

The present disclosure relates to an image processing method and apparatus, an electronic device, and a storage medium, the method including: performing feature extraction on an image to be processed to obtain a feature map of the image to be processed; performing first positioning and segmentation processing on the feature map, and determining a first segmentation result of a first target; performing second positioning and segmentation processing on the feature map, and determining a second segmentation result of a second target; and determining the segmentation result of the image to be processed according to the first segmentation result and the second segmentation result. The embodiment of the disclosure can realize the differentiation processing of the targets with different sizes in different areas in the image to be processed, and improve the precision of image processing.

Description

Image processing method and device, electronic equipment and storage medium

Technical Field

The present disclosure relates to the field of image processing technologies, and in particular, to an image processing method and apparatus, an electronic device, and a storage medium.

Background

In the field of image technology, segmentation of a region of interest or a target region is the basis for image analysis and target recognition. For example, boundaries between one or more organs or tissues are clearly identified by segmentation in medical images. Accurately segmenting medical images is crucial for many clinical applications.

Disclosure of Invention

The present disclosure proposes an image processing technical solution.

According to an aspect of the present disclosure, there is provided an image processing method including: performing feature extraction on an image to be processed to obtain a feature map of the image to be processed; performing first positioning and segmentation processing on the feature map, and determining a first segmentation result of a first target; performing second positioning and segmentation processing on the feature map, and determining a second segmentation result of a second target; and determining the segmentation result of the image to be processed according to the first segmentation result and the second segmentation result.

In a possible implementation manner, performing a second positioning and segmentation process on the feature map, and determining a second segmentation result of a second target includes: performing second positioning processing and cutting processing on the feature map to respectively obtain position information of a second target and a target feature map; and determining a second segmentation result of the second target according to the target feature map, the position information of the second target, the image to be processed and the feature map.

In a possible implementation manner, performing second positioning processing and cropping processing on the feature map to obtain position information of a second target and a target feature map, respectively, includes: performing second positioning processing on the feature map to determine position information of a second target; and performing cropping processing on the feature map according to the position information of the second target to obtain a target feature map of the second target.

In a possible implementation manner, determining a second segmentation result of the second target according to the target feature map, the position information of the second target, the image to be processed, and the feature map includes: performing image fusion on the target feature map, the position information of the second target, the image to be processed and the feature map to obtain a fusion result; and performing second segmentation on the fusion result, and determining a second segmentation result of the second target.

In a possible implementation manner, the feature map includes N layers of feature maps, where N is an integer greater than 1, where performing the second positioning process on the feature map to determine the location information of the second target includes: and carrying out second positioning processing on the N-th layer characteristic diagram to determine a position probability diagram of a second target.

In a possible implementation manner, the cropping the feature map according to the position information of the second target to obtain the target feature map of the second target includes: and performing cutting processing on the N-th layer of feature map according to the position information of the second target to obtain a target feature map of the second target.

In a possible implementation manner, image fusion is performed on the target feature map, the position information of the second target, the image to be processed, and the feature map, so as to obtain a fusion result, where the fusion result includes: according to the position information of the second target, respectively carrying out third segmentation on the image to be processed and the first layer characteristic graph to obtain a segmented image to be processed and a segmented first layer characteristic graph; and carrying out image fusion on the target feature map, the position information of the second target, the segmented image to be processed and the segmented first-layer feature map to obtain a fusion result.

In a possible implementation manner, performing feature extraction on an image to be processed to obtain a feature map of the image to be processed includes: carrying out convolution processing on an image to be processed to obtain a convolution result; carrying out residual error and compression activation processing on the convolution result to obtain an activation result; and performing multi-scale feature extraction and deconvolution processing on the activation result to obtain a feature map of the image to be processed.

In one possible implementation, the method is implemented by a neural network comprising a master segmentation network, a first positioning network and a first segmentation network, the master segmentation network comprising a feature extraction network and a second positioning and segmentation network,

the feature extraction network is used for extracting features of an image to be processed, the second positioning and segmentation network is used for performing first positioning and segmentation processing on the feature map, the first positioning network is used for performing second positioning processing on the feature map, and the first segmentation network is used for determining a second segmentation result of the second target.

In one possible implementation, the method further includes: and training the neural network according to a preset training set.

In one possible implementation, training the neural network according to a preset training set includes: training the main segmentation network according to the training set; training the first positioning network according to the training set and the trained main segmentation network; training the first segmentation network based on the training set, the trained primary segmentation network, and the trained first positioning network.

In one possible implementation, training the neural network according to a preset training set includes: determining the network loss of the neural network according to a focal loss function and a generalized distance function; and adjusting network parameters of the neural network according to the network loss.

In one possible implementation, the image to be processed is a medical image containing an organ at risk OAR.

According to an aspect of the present disclosure, there is provided an image processing apparatus including:

the characteristic extraction module is used for extracting the characteristics of the image to be processed to obtain a characteristic diagram of the image to be processed;

the first determination module is used for performing first positioning and segmentation processing on the feature map and determining a first segmentation result of a first target;

the second determining module is used for carrying out second positioning and segmentation processing on the feature map and determining a second segmentation result of a second target;

and the segmentation result determining module is used for determining the segmentation result of the image to be processed according to the first segmentation result and the second segmentation result.

In one possible implementation manner, the second determining module includes: the positioning sub-module is used for carrying out second positioning processing and cutting processing on the characteristic diagram to respectively obtain position information of a second target and a target characteristic diagram; and the determining submodule is used for determining a second segmentation result of the second target according to the target feature map, the position information of the second target, the image to be processed and the feature map.

In one possible implementation, the positioning sub-module is further configured to: performing second positioning processing on the feature map to determine position information of a second target; and performing cropping processing on the feature map according to the position information of the second target to obtain a target feature map of the second target.

In one possible implementation, the determining sub-module is further configured to: performing image fusion on the target feature map, the position information of the second target, the image to be processed and the feature map to obtain a fusion result; and performing second segmentation on the fusion result, and determining a second segmentation result of the second target.

In one possible implementation, the feature map includes N layers of feature maps, where N is an integer greater than 1, and the positioning sub-module is further configured to: and carrying out second positioning processing on the N-th layer characteristic diagram to determine a position probability diagram of a second target.

In one possible implementation, the positioning sub-module is further configured to: and performing cutting processing on the N-th layer of feature map according to the position information of the second target to obtain a target feature map of the second target.

In one possible implementation, the determining sub-module is further configured to: according to the position information of the second target, respectively carrying out third segmentation on the image to be processed and the first layer characteristic graph to obtain a segmented image to be processed and a segmented first layer characteristic graph; and carrying out image fusion on the target feature map, the position information of the second target, the segmented image to be processed and the segmented first-layer feature map to obtain a fusion result.

In one possible implementation, the feature extraction module includes: the convolution submodule is used for performing convolution processing on the image to be processed to obtain a convolution result; the activation submodule is used for carrying out residual error and compression activation processing on the convolution result to obtain an activation result; and the extraction submodule is used for carrying out multi-scale feature extraction and deconvolution processing on the activation result to obtain a feature map of the image to be processed.

In one possible implementation, the apparatus is implemented by a neural network comprising a master partition network, a first positioning network, and a first partition network, the master partition network comprising a feature extraction network and a second positioning and partition network,

In one possible implementation, the apparatus further includes: and the training module is used for training the neural network according to a preset training set.

In one possible implementation, the training module includes: the first training submodule is used for training the main segmentation network according to the training set; a second training submodule, configured to train the first positioning network according to the training set and the trained main segmentation network; and the third training submodule is used for training the first segmentation network according to the training set, the trained main segmentation network and the trained first positioning network.

In one possible implementation, the training module includes: the loss determining submodule is used for determining the network loss of the neural network according to a focal loss function and a generated variance loss function; and the adjusting submodule is used for adjusting the network parameters of the neural network according to the network loss.

According to an aspect of the present disclosure, there is provided an electronic device including:

a processor;

a memory for storing processor-executable instructions;

wherein the processor is configured to: the above-described image processing method is performed.

According to an aspect of the present disclosure, there is provided a computer-readable storage medium having stored thereon computer program instructions which, when executed by a processor, implement the above-described image processing method.

In the embodiment of the disclosure, for a first target and a second target, a first segmentation result of the first target is obtained by performing first positioning and segmentation processing on an extracted feature map of an image to be processed, a segmentation result of the second target is obtained by performing second positioning and segmentation processing on the feature map, and a segmentation result of the image to be processed is obtained according to the segmentation result of the first target and the segmentation result of the second target. By the process, the differentiation processing of the targets with different sizes in different areas in the image to be processed can be realized, and the precision of image processing is improved.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. Other features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments, which proceeds with reference to the accompanying drawings.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the disclosure.

Fig. 1 shows a flowchart of an image processing method according to an embodiment of the present disclosure.

Fig. 2a shows a flowchart of the method of step S10 according to an embodiment of the present disclosure.

Fig. 2b shows a flowchart of the method of step S12 according to an embodiment of the present disclosure.

Fig. 3 shows a schematic diagram of a neural network according to an embodiment of the present disclosure.

Fig. 4 shows a flowchart of the method of step S122 according to an embodiment of the present disclosure.

Fig. 5 shows a flowchart of an image processing method according to an embodiment of the present disclosure.

Fig. 6 shows a flowchart of the method of step S14 according to an embodiment of the present disclosure.

Fig. 7 illustrates a block diagram of an image processing apparatus according to an embodiment of the present disclosure.

FIG. 8 shows a block diagram of an electronic device in accordance with an embodiment of the disclosure.

Fig. 9 shows a block diagram of an electronic device in accordance with an embodiment of the disclosure.

Detailed Description

Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. In the drawings, like reference numbers can indicate functionally identical or similar elements. While the various aspects of the embodiments are presented in drawings, the drawings are not necessarily drawn to scale unless specifically indicated.

The word "exemplary" is used exclusively herein to mean "serving as an example, embodiment, or illustration. Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments.

The term "and/or" herein is merely an association describing an associated object, meaning that three relationships may exist, e.g., a and/or B, may mean: a exists alone, A and B exist simultaneously, and B exists alone. In addition, the term "at least one" herein means any one of a plurality or any combination of at least two of a plurality, for example, including at least one of A, B, C, and may mean including any one or more elements selected from the group consisting of A, B and C.

Furthermore, in the following detailed description, numerous specific details are set forth in order to provide a better understanding of the present disclosure. It will be understood by those skilled in the art that the present disclosure may be practiced without some of these specific details. In some instances, methods, means, elements and circuits that are well known to those skilled in the art have not been described in detail so as not to obscure the present disclosure.

Fig. 1 shows a flowchart of an image processing method according to an embodiment of the present disclosure, which may be applied to an image processing apparatus, which may be a terminal device, a server, or other processing device, and the like. The terminal device may be a User Equipment (UE), a mobile device, a User terminal, a cellular phone, a cordless phone, a Personal Digital Assistant (PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, or the like.

In some possible implementations, the image processing method may be implemented by a processor calling computer readable instructions stored in a memory.

As shown in fig. 1, the image processing method may include:

step S10, extracting the characteristics of the image to be processed to obtain the characteristic diagram of the image to be processed;

step S11, performing first positioning and segmentation processing on the feature map, and determining a first segmentation result of the first target;

step S12, performing second positioning and segmentation processing on the feature map, and determining a second segmentation result of a second target;

step S13, determining a segmentation result of the image to be processed according to the first segmentation result and the second segmentation result.

The size of the first target may be larger than that of the second target, or the size of the first target may also be smaller than that of the second target, which is not limited in the present disclosure.

According to the image processing method, for a first target and a second target with different sizes, a first segmentation result of the first target is obtained by performing first positioning and segmentation processing on the extracted feature map of the image to be processed, a segmentation result of the second target is obtained by performing second positioning and segmentation processing on the feature map, and a segmentation result of the image to be processed is obtained according to the segmentation result of the first target and the segmentation result of the second target. By the process, the differentiation processing of the targets with different sizes in different areas in the image to be processed can be realized, and the precision of image processing is improved.

The image processing method disclosed by the embodiment of the disclosure can realize automatic and efficient segmentation of the target object in the image, and can segment the targets with different sizes in the image to be processed through different segmentation processes, so as to obtain a more accurate segmentation result.

The image processing method of the embodiment of the present disclosure may be applied to processing of medical images, for example, to identify a target region in a medical image, where the target region may be a lesion, a diseased organ, an organ at risk, and the like. In a possible implementation manner, the image to be processed may be a medical image including an organ at risk OAR, that is, the image processing method of the embodiment of the disclosure may be applied to a clinical radiotherapy planning process for identifying an organ at risk, and by accurately identifying the position of the organ at risk, side effects of radiotherapy on normal organs are reduced, and the effect of radiotherapy is improved.

It should be noted that the image processing method according to the embodiment of the present disclosure is not limited to be applied to medical image processing, and may be applied to any image processing, and the present disclosure does not limit this.

In one possible implementation, the image to be processed may include a plurality of pictures from which one or more three-dimensional organs may be identified.

For step S10, a related feature extraction technique may be employed to extract a feature map of the image to be processed. For example, the feature map of the image to be processed may be extracted based on the artificially designed features such as the local brightness feature of the image, the shape feature of the organ, and the like.

In a possible implementation manner, a 3D U-Net full convolution neural network based on an Encoder-Decoder (Encoder-Decoder) architecture may be adopted to perform one or more times of convolution processing on an image to be processed to obtain a convolution result, and then perform a corresponding number of times of deconvolution processing to obtain a feature map of the image to be processed. The present disclosure does not limit the specific manner of feature extraction.

In another possible implementation manner, fig. 2a shows a flowchart of the method of step S10 according to an embodiment of the present disclosure, and as shown in fig. 2a, step S10 may include:

step S101, performing convolution processing on an image to be processed to obtain a convolution result;

step S102, carrying out residual error and compression activation processing on the convolution result to obtain an activation result;

and S103, performing multi-scale feature extraction and deconvolution processing on the activation result to obtain a feature map of the image to be processed.

In one possible implementation, the image processing method of the embodiments of the present disclosure may be implemented by a neural network. Fig. 3 shows a schematic diagram of a neural network according to an embodiment of the present disclosure, as shown in fig. 3, the neural network may include a main segmentation network 1 including a feature extraction network and a second positioning and segmentation network 13, a first positioning network 2, and a first segmentation network 3.

The feature extraction network performs feature extraction on an image to be processed to obtain a feature map 12 of the image to be processed, the second positioning and segmentation network 13 is configured to perform first positioning and segmentation on the feature map, the first positioning network 2 is configured to perform second positioning on the feature map 12, and the first segmentation network 3 is configured to determine a second segmentation result 31 of the second target.

According to the neural network of the embodiment of the present disclosure, the image processing method is implemented, parameters (for example, feature maps and the like) are shared among the main segmentation network 1, the first positioning network 2 and the first segmentation network 3, redundant calculation is not performed, and the segmentation efficiency and the segmentation accuracy are improved.

In one possible implementation, the main segmentation network 1 may employ a convolutional neural network modified based on the 3D U-Net of the Encoder-Decoder architecture, and the main segmentation network 1 may include a feature extraction network and the second positioning and segmentation network 13. The feature extraction network may include a Residual and compression activation module, Squeeze-and-Excitation Residual Block (SEResBlock), and a densely connected void convolutional spatial pyramid module (denseas pp).

The feature extraction network can reduce the down-sampling times of the image to be processed, thereby reducing the loss of high-resolution information; meanwhile, in order to enhance the feature expression capability of the network, the feature extraction network uses a residual module (Residue Block, including a convolutional layer, a linear rectification function, and a batch normalization layer) as a basic structure, further adds a compression-activation module (SE module) as an attention mechanism of a feature layer, captures and learns features of different scales through denoaspp to fuse multi-scale features and ensure a sufficiently large convolutional kernel perception field (perceptual field), and can realize the learning of the features of different scales by setting the expansion rate (convolution) of the convolution.

For the above steps S101 to S103, the feature extraction network may perform convolution processing on the image to be processed to obtain a convolution result, where the feature extraction network may include N layers of convolution layers, where N is an integer greater than 1. And then carrying out residual error and compression activation processing on the convolution result through a residual error and compression activation module to obtain an activation result. And then, performing multi-scale feature extraction on the activation result through DenseASPP, and then performing deconvolution processing to obtain a feature map of the image to be processed.

It should be noted that the above embodiments are only some examples of the disclosure, and do not limit the disclosure in any way. It can be understood by those skilled in the art that feature extraction of the image to be processed can also be implemented in other ways as long as the feature map of the image to be processed can be obtained.

The first target may be a larger body organ than the second target, for example the first target may be the parotid gland, the second target may be the lens of the eye, etc. With respect to step S11, by further performing the first positioning and segmentation process on the feature map by the above-described main segmentation network 1, a first segmentation result of the first target may be determined, for example, an amplifier in the image to be processed may be determined. The embodiments of the present disclosure do not limit the specific procedures and the specific techniques used in the first positioning and segmentation process, and are not limited to the method using a neural network.

For step S12, a second positioning and segmentation process may be performed on the feature map through the first positioning network 2 and the first segmentation network 3, and a second segmentation result of the second target may be determined, for example, a small organ in the image to be processed may be determined. The specific process and the specific technique adopted in the second positioning and segmentation process are not limited, and are not limited to the method by the neural network.

Fig. 2b shows a flowchart of the method of step S12 according to the embodiment of the present disclosure, and as shown in fig. 2b, in one possible implementation, step S12 may include:

step S121, performing second positioning processing and cutting processing on the feature map to respectively obtain position information of a second target and a target feature map;

and step S122, determining a second segmentation result of the second target according to the target feature map, the position information of the second target, the image to be processed and the feature map.

For step S121, a second positioning process may be performed on the feature map to determine position information of a second target; and performing cropping processing on the feature map according to the position information of the second target to obtain a target feature map of the second target. Then, in step S122, the target feature map of the second target, the position information of the second target, the image to be processed, and the feature map are integrated to further precisely segment the second target.

As described above, the feature extraction network may include N convolutional layers, where N is an integer greater than 1, and the feature map may include N feature maps, where the "performing the second positioning process on the feature map to determine the location information of the second target" may include: and carrying out second positioning processing on the N-th layer characteristic diagram to determine a position probability diagram of a second target.

It should be noted that, the second positioning processing may also be performed on feature maps of other layers at the same time (for example, the second positioning processing is performed on feature maps of the first layer and the last layer at the same time), and a position probability map of the second object is determined, so as to obtain a more accurate second object, which is not limited in this disclosure.

In a possible implementation manner, as shown in fig. 3, the neural network may further include a first positioning network 2, and the first positioning network 2 may include two seresblocks. The N-th layer of feature map (the last layer of feature map of the decoder of the feature extraction network) obtained by feature extraction of the image to be processed by the feature extraction network is input to the first positioning network 2, and the first positioning network 2 can perform second positioning processing on the last layer of feature map to determine the position probability map of the second target. Specifically, the first positioning network 2 may first position the center position of the second target, and create a 3D gaussian distribution map of the center position of the second target as the position probability map of the second target.

In a possible implementation manner, for second targets with different sizes or shapes, the first positioning network 2 corresponding to the second target may be separately set to position the second target. That is, the neural network of the embodiment of the present disclosure may include a plurality of first positioning networks 2.

It should be noted that the positioning manner of the second target is not limited to the above example, and those skilled in the art can understand that the second target may also be positioned by other techniques to obtain the position information of the second target, for example, the position of the second target is obtained by performing image registration based on a atlas method, or a detection frame containing the second target is obtained by a target detection method.

The above "performing a cropping process on the feature map according to the position information of the second object to obtain the object feature map of the second object" may include: and performing cutting processing on the N-th layer of feature map according to the position information of the second target to obtain a target feature map of the second target.

The feature map of the nth layer may also be the feature map of the last layer of the feature extraction network decoder as described above, where the feature map of the nth layer includes the multi-scale feature quantity. After the position information of the second target is obtained, the nth layer of feature map may be cropped according to the position information of the second target to obtain a target feature map of the second target. In other words, the feature portion of the second object in the feature map is cut out, for example, a portion composed of pixels at a corresponding position is found in the feature map according to the position information of the second object as the object feature map of the second object.

In a possible implementation manner, the feature map of the nth layer may be cropped according to the position probability map of the second target to obtain a segmented feature map of the second target.

It should be noted that, feature maps of other layers may also be cut at the same time to obtain a target feature map of the second target, so as to obtain a more accurate second target, which is not limited in this disclosure.

For step S122, the target feature map includes more scale feature quantities, the image to be processed includes high resolution feature quantities, and a second segmentation result of the second target is determined by integrating the target feature map, the position information of the second target, the image to be processed, and the feature map, so that accurate segmentation of the second target with a smaller size can be achieved.

Fig. 4 shows a flowchart of the method of step S122 according to an embodiment of the present disclosure, and as shown in fig. 4, determining a second segmentation result of the second target according to the target feature map, the position information of the second target, the image to be processed, and the feature map in step S122 may include:

step S1221, carrying out image fusion on the target feature map, the position information of the second target, the image to be processed and the feature map to obtain a fusion result;

step S1222, performing a second segmentation on the fusion result, and determining a second segmentation result of the second target.

By fusing the target feature map, the position information of the second target, the image to be processed and the multi-aspect information of the feature map, more image information (such as scale, feature, resolution and the like) can be obtained, the precision of subsequent image segmentation is improved, and the accurate segmentation of the target with smaller size is facilitated. The process of image fusion can be realized by using a related image fusion system, and the embodiment of the disclosure does not limit the specific fusion process.

In a possible implementation manner, before the fusion, roi (region of interest) pooling may be performed on the feature map, the image to be processed, and the position information of the second target, so as to reduce data dimension and improve processing efficiency.

In a possible implementation manner, as shown in fig. 3, the neural network may further include a first segmentation network 3, where the first segmentation network 3 may also be built by SEResBlock, for example, in an example, the first segmentation network 3 may include 5 seresblocks. And performing second segmentation processing on the fusion result through the first segmentation network, so that an accurate second segmentation result of the second target can be obtained.

In a possible implementation manner, for second targets with different sizes or shapes, the first segmentation network 3 corresponding to the second target can be separately set to segment the second target. That is, the neural network of the embodiment of the present disclosure may include a plurality of first divided networks 3.

In one possible implementation, step S1221 may include: according to the position information of the second target, respectively carrying out third segmentation on the image to be processed and the first layer characteristic graph to obtain a segmented image to be processed and a segmented first layer characteristic graph; and carrying out image fusion on the target feature map, the position information of the second target, the segmented image to be processed and the segmented first-layer feature map to obtain a fusion result.

The image to be processed and the first layer feature map include high-resolution feature quantities, the image to be processed and the first layer feature map are further segmented through the position information of the second target, the position information of the second target is obtained by performing second positioning processing on the feature map extracted by the feature extraction network, and the segmentation result of the second target is coded in the target feature map, so that the segmented image to be processed and the segmented first layer feature map are obtained, and image fusion is performed on the basis of the segmented image to be processed and the segmented first layer feature map, so that the segmentation accuracy can be further improved.

For step S13, as shown in fig. 3, in order to output the uniform segmentation results of all the targets, the first segmentation result of the first target and the second segmentation result of the second target may be fused to obtain the segmentation result of the image to be processed for outputting the final segmentation map 4.

Fig. 5 shows a flowchart of an image processing method according to an embodiment of the present disclosure. In one possible implementation manner, as shown in fig. 5, the method of the embodiment of the present disclosure may further include:

and step S14, training the neural network according to a preset training set.

The preset training set may be a plurality of picture sets obtained by splitting a sample picture after preprocessing such as manual clipping. The method includes splitting the image into a plurality of image sets, where two adjacent image sets may include a part of the same image, for example, in the case of a medical image, a plurality of samples may be taken from a hospital, a plurality of sample images included in one sample may be continuously taken of an organ of a human body, a three-dimensional stereoscopic structure of the organ may be obtained by the plurality of sample images, the splitting may be performed along one direction, a first image set may include 1 st to 30 th images, and a second image set may include 16 th to 45 th images … …, so that 15 images in two adjacent image sets are the same. By means of the overlapping splitting, the accuracy of segmentation can be improved.

Fig. 6 illustrates a flowchart of the method of step S14 according to an embodiment of the present disclosure, and as shown in fig. 6, step S14 may include:

step S141, training the main segmentation network according to the training set;

step S142, training the first positioning network according to the training set and the trained main segmentation network;

step S143, training the first segmentation network according to the training set, the trained primary segmentation network, and the trained first positioning network.

As shown in fig. 3, the main segmentation network is trained, and then the first positioning network is trained according to the training set and the main segmentation network under the condition that parameters of the main segmentation network are fixed, that is, the training set is input into the trained main segmentation network, and a feature diagram obtained by extracting features of the training set by the trained main segmentation network is input into the first positioning network to train the first positioning network.

And finally, training the first positioning network according to the trained main segmentation network, the trained first positioning network and the training set. Specifically, a training set is input into a trained main segmentation network, the trained main segmentation network extracts features of the training set to obtain a feature map, the trained first positioning network is used for positioning the feature map to obtain position information of a target, the feature map is cut according to the position information of the target to obtain a target feature map, and the target feature map, the position information of the target, the training set and the feature map are input into the first segmentation network to train the first segmentation network.

In a possible implementation manner, during the training process, the network loss of the neural network can be determined according to a focus loss function and a generalized dice loss function; and adjusting network parameters of the neural network according to the network loss. That is, training the neural network according to a preset training set may further include: the above procedure adjusts the network parameters according to the network loss.

For example, as shown in fig. 3, the main segmentation network may be trained according to a training set, and during the training of the main segmentation network, the network loss of the main segmentation network may be determined according to a local loss function and a generalized distance function; and adjusting the network parameters of the main segmentation network according to the network loss until the network loss meets a preset condition, for example, the network loss does not decrease any more.

In one possible implementation, the network loss of the primary split network may be determined according to the following equation (1):

L_total＝L_Focal+λL_Dice (1)

wherein L is_totalFor total network loss, L_FocalIs focal length, L_DiceFor generated loss, λ is a weight that balances the proportion of the two losses in the total loss, and in one example, λ may be 1.

After the training of the main segmentation network is completed, the first positioning network can be trained according to the training set and the trained main segmentation network, and in the process of training the first positioning network, the network Loss of the first positioning network can be determined according to an MSE (mean square error Loss) function; and adjusting the network parameters of the first positioning network according to the network loss until the network loss meets the preset condition.

After the training of the first positioning network is completed, the first segmentation network can be trained according to the training set, the trained main segmentation network and the trained first positioning network, and the network loss of the first segmentation network can still be determined according to the focus loss function and the generalized dice loss function in the process of training the first segmentation network; and adjusting the network parameters of the first segmentation network according to the network loss until the network loss meets the preset condition.

In one possible implementation, the network loss of the first split network may also be determined according to equation (1) above.

It should be noted that, in the process of training the main segmentation network and the first segmentation network, the parameters of the focus loss function and the generalized die loss function may be the same or different, and the parameters of the focus loss function and/or the generalized die loss function may be selected according to the characteristics of the networks, which is not limited in the embodiment of the present disclosure.

According to the embodiment of the disclosure, the neural network is trained, and the main segmentation network, the first positioning network and the first segmentation network are respectively trained, so that the trained neural network can identify targets with different sizes even if samples of a training set are unbalanced, and the segmentation accuracy of the targets with different sizes in an image is improved.

In addition, the image processing method of the embodiment of the disclosure adopts a local loss function and a generalized difference loss function to evaluate the network loss of the neural network, and in the training process, the parameters of the neural network are adjusted according to the network loss, so that the influence of sample imbalance on the determined network loss is reduced, the problem caused by sample imbalance is further solved, the training effect is improved, the neural network obtained by training is more suitable for recognizing targets with different sizes, and the segmentation accuracy of the targets with different sizes in the image is improved.

Application scenario example

Clinically, when planning radiotherapy, over twenty Organs At Risk (OAR) need to be taken into account, and a doctor is usually required to delineate on a three-dimensional Computed Tomography (CT) image, however, labeling on a three-dimensional CT image is usually time-consuming and labor-consuming.

For example, due to the complicated anatomical structure of the head and neck and the characteristic that CT is insensitive to soft tissues, many organs have low contrast with surrounding tissues and unclear boundaries, which further increases the difficulty of delineation and puts high requirements on the professional performance of doctors. It usually takes a professional physician over 2.5 hours to delineate one patient's organ, and in addition, due to subjective factors, delineations of the same patient's organ by different physicians may not be exactly consistent.

Therefore, the computer-aided segmentation method with high speed, high efficiency, excellent performance and strong robustness can greatly reduce the workload of doctors, improve the speed and the quality of radiotherapy planning and improve the effect of radiotherapy.

Organs at risk contain many organs of varying volumes and, therefore, present a significant sample imbalance problem. For large organs, such as the parotid gland, the volume is more than 250 times that of the smallest organ lens. How to balance large organs and small organs and have better segmentation precision on different organs is a problem which needs to be solved urgently.

By adopting the image processing method disclosed by the invention to process the medical image containing a plurality of organs with different volumes, the accurate segmentation of organs with different volumes can be realized, especially the accurate segmentation of small organs can be realized, the radiotherapy planning speed and quality are improved, and the radiotherapy effect is improved.

It is understood that the above-mentioned method embodiments of the present disclosure can be combined with each other to form a combined embodiment without departing from the logic of the principle, which is limited by the space, and the detailed description of the present disclosure is omitted.

It will be understood by those skilled in the art that in the method of the present invention, the order of writing the steps does not imply a strict order of execution and any limitations on the implementation, and the specific order of execution of the steps should be determined by their function and possible inherent logic.

Fig. 7 illustrates a block diagram of an image processing apparatus according to an embodiment of the present disclosure. The image processing apparatus may be a terminal device, a server or other processing device, etc. The terminal device may be a User Equipment (UE), a mobile device, a User terminal, a cellular phone, a cordless phone, a Personal Digital Assistant (PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, or the like.

In some possible implementations, the image processing apparatus may be implemented by a processor calling computer readable instructions stored in a memory.

As shown in fig. 7, the image processing apparatus may include:

the feature extraction module 71 is configured to perform feature extraction on an image to be processed to obtain a feature map of the image to be processed;

a first determining module 72, configured to perform first positioning and segmentation processing on the feature map, and determine a first segmentation result of the first target;

a second determining module 73, configured to perform second positioning and segmentation processing on the feature map, and determine a second segmentation result of a second target;

and a segmentation result determining module 74, configured to determine a segmentation result of the image to be processed according to the first segmentation result and the second segmentation result.

Aiming at the first target and the second target, respectively, performing first positioning and segmentation processing on the extracted feature map of the image to be processed to obtain a first segmentation result of the first target, performing second positioning and segmentation processing on the feature map to obtain a segmentation result of the second target, and obtaining the segmentation result of the image to be processed according to the segmentation result of the first target and the segmentation result of the second target. By the process, the differentiation processing of the targets with different sizes in different areas in the image to be processed can be realized, and the precision of image processing is improved.

In one possible implementation, the second determining module 73 includes:

the positioning sub-module is used for carrying out second positioning processing and cutting processing on the characteristic diagram to respectively obtain position information of a second target and a target characteristic diagram;

and the determining submodule is used for determining a second segmentation result of the second target according to the target feature map, the position information of the second target, the image to be processed and the feature map.

In one possible implementation, the positioning sub-module is further configured to:

performing second positioning processing on the feature map to determine position information of a second target;

and performing cropping processing on the feature map according to the position information of the second target to obtain a target feature map of the second target.

In one possible implementation, the determining sub-module is further configured to:

performing image fusion on the target feature map, the position information of the second target, the image to be processed and the feature map to obtain a fusion result;

and performing second segmentation on the fusion result, and determining a second segmentation result of the second target.

and performing cutting processing on the N-th layer of feature map according to the position information of the second target to obtain a target feature map of the second target.

according to the position information of the second target, respectively carrying out third segmentation on the image to be processed and the first layer characteristic graph to obtain a segmented image to be processed and a segmented first layer characteristic graph;

and carrying out image fusion on the target feature map, the position information of the second target, the segmented image to be processed and the segmented first-layer feature map to obtain a fusion result.

In one possible implementation, the feature extraction module 71 includes:

the convolution submodule is used for performing convolution processing on the image to be processed to obtain a convolution result;

the activation submodule is used for carrying out residual error and compression activation processing on the convolution result to obtain an activation result;

and the extraction submodule is used for carrying out multi-scale feature extraction and deconvolution processing on the activation result to obtain a feature map of the image to be processed.

In one possible implementation, the apparatus further includes:

and a training module 75, configured to train the neural network according to a preset training set.

In one possible implementation, the training module 75 includes:

the first training submodule is used for training the main segmentation network according to the training set;

a second training submodule, configured to train the first positioning network according to the training set and the trained main segmentation network;

and the third training submodule is used for training the first segmentation network according to the training set, the trained main segmentation network and the trained first positioning network.

In one possible implementation, the training module 75 includes:

the loss determining submodule is used for determining the network loss of the neural network according to a focal loss function and a generated variance loss function;

and the adjusting submodule is used for adjusting the network parameters of the neural network according to the network loss.

Embodiments of the present disclosure also provide a computer-readable storage medium having stored thereon computer program instructions, which when executed by a processor, implement the above-mentioned method. The computer readable storage medium may be a non-volatile computer readable storage medium.

An embodiment of the present disclosure further provides an electronic device, including: a processor; a memory for storing processor-executable instructions; wherein the processor is configured as the above method.

The electronic device may be provided as a terminal, server, or other form of device.

Fig. 8 is a block diagram of an electronic device 800 in accordance with an embodiment of the disclosure. For example, the electronic device 800 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, or the like terminal.

Referring to fig. 8, electronic device 800 may include one or more of the following components: processing component 802, memory 804, power component 806, multimedia component 808, audio component 810, input/output (I/O) interface 812, sensor component 814, and communication component 816.

The processing component 802 generally controls overall operation of the electronic device 800, such as operations associated with display, telephone calls, data communications, camera operations, and recording operations. The processing components 802 may include one or more processors 820 to execute instructions to perform all or a portion of the steps of the methods described above. Further, the processing component 802 can include one or more modules that facilitate interaction between the processing component 802 and other components. For example, the processing component 802 can include a multimedia module to facilitate interaction between the multimedia component 808 and the processing component 802.

The memory 804 is configured to store various types of data to support operations at the electronic device 800. Examples of such data include instructions for any application or method operating on the electronic device 800, contact data, phonebook data, messages, pictures, videos, and so forth. The memory 804 may be implemented by any type or combination of volatile or non-volatile memory devices such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disks.

The power supply component 806 provides power to the various components of the electronic device 800. The power components 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for the electronic device 800.

The multimedia component 808 includes a screen that provides an output interface between the electronic device 800 and a user. In some embodiments, the screen may include a Liquid Crystal Display (LCD) and a Touch Panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive an input signal from a user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundary of a touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front facing camera and/or a rear facing camera. The front camera and/or the rear camera may receive external multimedia data when the electronic device 800 is in an operation mode, such as a shooting mode or a video mode. Each front camera and rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.

The audio component 810 is configured to output and/or input audio signals. For example, the audio component 810 includes a Microphone (MIC) configured to receive external audio signals when the electronic device 800 is in an operational mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signals may further be stored in the memory 804 or transmitted via the communication component 816. In some embodiments, audio component 810 also includes a speaker for outputting audio signals.

The I/O interface 812 provides an interface between the processing component 802 and peripheral interface modules, which may be keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to: a home button, a volume button, a start button, and a lock button.

The sensor assembly 814 includes one or more sensors for providing various aspects of state assessment for the electronic device 800. For example, the sensor assembly 814 may detect an open/closed state of the electronic device 800, the relative positioning of components, such as a display and keypad of the electronic device 800, the sensor assembly 814 may also detect a change in the position of the electronic device 800 or a component of the electronic device 800, the presence or absence of user contact with the electronic device 800, orientation or acceleration/deceleration of the electronic device 800, and a change in the temperature of the electronic device 800. Sensor assembly 814 may include a proximity sensor configured to detect the presence of a nearby object without any physical contact. The sensor assembly 814 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 814 may also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

The communication component 816 is configured to facilitate wired or wireless communication between the electronic device 800 and other devices. The electronic device 800 may access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 816 receives a broadcast signal or broadcast related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 further includes a Near Field Communication (NFC) module to facilitate short-range communications. For example, the NFC module may be implemented based on Radio Frequency Identification (RFID) technology, infrared data association (IrDA) technology, Ultra Wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

In an exemplary embodiment, the electronic device 800 may be implemented by one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), controllers, micro-controllers, microprocessors or other electronic components for performing the above-described methods.

In an exemplary embodiment, a non-transitory computer-readable storage medium, such as the memory 804, is also provided that includes computer program instructions executable by the processor 820 of the electronic device 800 to perform the above-described methods.

Fig. 9 is a block diagram of an electronic device 1900 according to an embodiment of the disclosure. For example, the electronic device 1900 may be provided as a server. Referring to fig. 9, electronic device 1900 includes a processing component 1922 further including one or more processors and memory resources, represented by memory 1932, for storing instructions, e.g., applications, executable by processing component 1922. The application programs stored in memory 1932 may include one or more modules that each correspond to a set of instructions. Further, the processing component 1922 is configured to execute instructions to perform the above-described method.

The electronic device 1900 may also include a power component 1926 configured to perform power management of the electronic device 1900, a wired or wireless network interface 1950 configured to connect the electronic device 1900 to a network, and an input/output (I/O) interface 1958. The electronic device 1900 may operate based on an operating system stored in memory 1932, such as Windows Server, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, or the like.

In an exemplary embodiment, a non-transitory computer readable storage medium, such as the memory 1932, is also provided that includes computer program instructions executable by the processing component 1922 of the electronic device 1900 to perform the above-described methods.

The present disclosure may be systems, methods, and/or computer program products. The computer program product may include a computer-readable storage medium having computer-readable program instructions embodied thereon for causing a processor to implement various aspects of the present disclosure.

The computer readable storage medium may be a tangible device that can hold and store the instructions for use by the instruction execution device. The computer readable storage medium may be, for example, but not limited to, an electronic memory device, a magnetic memory device, an optical memory device, an electromagnetic memory device, a semiconductor memory device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a Static Random Access Memory (SRAM), a portable compact disc read-only memory (CD-ROM), a Digital Versatile Disc (DVD), a memory stick, a floppy disk, a mechanical coding device, such as punch cards or in-groove projection structures having instructions stored thereon, and any suitable combination of the foregoing. Computer-readable storage media as used herein is not to be construed as transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission medium (e.g., optical pulses through a fiber optic cable), or electrical signals transmitted through electrical wires.

The computer-readable program instructions described herein may be downloaded from a computer-readable storage medium to a respective computing/processing device, or to an external computer or external storage device via a network, such as the internet, a local area network, a wide area network, and/or a wireless network. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers and/or edge servers. The network adapter card or network interface in each computing/processing device receives computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in the respective computing/processing device.

The computer program instructions for carrying out operations of the present disclosure may be assembler instructions, Instruction Set Architecture (ISA) instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider). In some embodiments, the electronic circuitry that can execute the computer-readable program instructions implements aspects of the present disclosure by utilizing the state information of the computer-readable program instructions to personalize the electronic circuitry, such as a programmable logic circuit, a Field Programmable Gate Array (FPGA), or a Programmable Logic Array (PLA).

Various aspects of the present disclosure are described herein with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the disclosure. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer-readable program instructions.

These computer-readable program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. These computer-readable program instructions may also be stored in a computer-readable storage medium that can direct a computer, programmable data processing apparatus, and/or other devices to function in a particular manner, such that the computer-readable medium storing the instructions comprises an article of manufacture including instructions which implement the function/act specified in the flowchart and/or block diagram block or blocks.

The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer implemented process such that the instructions which execute on the computer, other programmable apparatus or other devices implement the functions/acts specified in the flowchart and/or block diagram block or blocks.

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

Having described embodiments of the present disclosure, the foregoing description is intended to be exemplary, not exhaustive, and not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The terms used herein were chosen in order to best explain the principles of the embodiments, the practical application, or technical improvements to the techniques in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. An image processing method, comprising:

performing feature extraction on an image to be processed to obtain a feature map of the image to be processed;

performing first positioning and segmentation processing on the feature map, and determining a first segmentation result of a first target;

performing second positioning and segmentation processing on the feature map, and determining a second segmentation result of a second target;

determining a segmentation result of the image to be processed according to the first segmentation result and the second segmentation result;

performing second positioning and segmentation processing on the feature map, and determining a second segmentation result of a second target, wherein the second positioning and segmentation processing comprises the following steps:

performing second positioning processing and cutting processing on the feature map to respectively obtain position information of a second target and a target feature map;

determining the second segmentation result according to the target feature map, the position information of the second target, the image to be processed and the feature map;

determining a second segmentation result of the second target according to the target feature map, the position information of the second target, the image to be processed and the feature map, wherein the second segmentation result comprises the following steps:

performing second segmentation on the fusion result, and determining a second segmentation result of the second target;

the feature map includes N layers of feature maps, where N is an integer greater than 1, and the target feature map, the position information of the second target, the image to be processed, and the feature map are subjected to image fusion to obtain a fusion result, where the fusion result includes:

2. The method according to claim 1, wherein performing a second positioning process and a cropping process on the feature map to obtain position information of a second target and a target feature map respectively comprises:

3. The method of claim 2, wherein the feature map comprises N layers of feature maps, N being an integer greater than 1,

performing second positioning processing on the feature map to determine position information of a second target, including:

and carrying out second positioning processing on the N-th layer characteristic diagram to determine a position probability diagram of a second target.

4. The method according to claim 3, wherein performing a cropping process on the feature map according to the position information of the second object to obtain an object feature map of the second object comprises:

5. The method according to any one of claims 1 to 4, wherein the performing feature extraction on the image to be processed to obtain the feature map of the image to be processed comprises:

carrying out convolution processing on an image to be processed to obtain a convolution result;

carrying out residual error and compression activation processing on the convolution result to obtain an activation result;

and performing multi-scale feature extraction and deconvolution processing on the activation result to obtain a feature map of the image to be processed.

6. The method according to any one of claims 2-4, wherein the method is implemented by a neural network comprising a master segmentation network, a first positioning network and a first segmentation network, the master segmentation network comprising a feature extraction network and a second positioning and segmentation network,

7. The method of claim 6, further comprising:

and training the neural network according to a preset training set.

8. The method of claim 7, wherein training the neural network according to a preset training set comprises:

training the main segmentation network according to the training set;

training the first positioning network according to the training set and the trained main segmentation network;

training the first segmentation network based on the training set, the trained primary segmentation network, and the trained first positioning network.

9. The method of claim 7 or 8, wherein training the neural network according to a preset training set comprises:

determining the network loss of the neural network according to a focal loss function and a generalized distance function;

and adjusting network parameters of the neural network according to the network loss.

10. The method according to any of claims 1-4, wherein the image to be processed is a medical image containing an Organ At Risk (OAR).

11. An image processing apparatus characterized by comprising: the characteristic extraction module is used for extracting the characteristics of the image to be processed to obtain a characteristic diagram of the image to be processed;

the segmentation result determining module is used for determining the segmentation result of the image to be processed according to the first segmentation result and the second segmentation result;

wherein the second determining module comprises:

the determining submodule is used for determining the second segmentation result according to the target feature map, the position information of the second target, the image to be processed and the feature map;

the determination submodule is further configured to:

12. The apparatus of claim 11, wherein the positioning sub-module is further configured to:

13. The apparatus of claim 12, wherein the positioning sub-module is further configured to:

14. The apparatus of claim 13, wherein the positioning sub-module is further configured to:

15. The apparatus according to any one of claims 11-14, wherein the feature extraction module comprises:

16. The apparatus according to any one of claims 12-14, wherein the apparatus is implemented by a neural network comprising a master segmentation network, a first positioning network and a first segmentation network, the master segmentation network comprising a feature extraction network and a second positioning and segmentation network,

17. The apparatus of claim 16, further comprising:

and the training module is used for training the neural network according to a preset training set.

18. The apparatus of claim 17, wherein the training module comprises:

19. The apparatus of claim 17 or 18, wherein the training module comprises:

20. The apparatus according to any of claims 11-14, wherein the image to be processed is a medical image containing an organ at risk OAR.

21. An electronic device, comprising:

a processor;

a memory for storing processor-executable instructions;

wherein the processor is configured to: performing the method of any one of claims 1 to 10.

22. A computer readable storage medium having computer program instructions stored thereon, which when executed by a processor implement the method of any one of claims 1 to 10.