Image segmentation method and device, electronic device, and storage medium

Abstract

The application discloses an image segmentation method and device, an electronic device and a storage medium. The method comprises the following steps: acquiring a first medical image; converting the first medical image into a second medical image of a target type, wherein the target type comprises at least one of a target phase, a target contrast, a target mode and a target resolution; inputting the second medical image into an image segmentation model to perform image segmentation processing on a region of interest in the second medical image; wherein the image segmentation model is obtained by training a neural network by using image samples of the target type, and the image samples are labeled with identification information of the region of interest; and determining a segmentation result of the first medical image according to a segmentation result of the second medical image. Therefore, a plurality of image segmentation models of different types do not need to be trained, and image samples used for training the image segmentation models of different types do not need to be labeled, so that a large amount of human cost can be saved.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to an image segmentation method and apparatus, electronic device, and storage medium. Background Technology

[0002] In the field of neural networks, training samples are typically used to train neural networks to obtain models with the expected functions, such as image segmentation models. The quantity and richness of training samples directly affect the accuracy of the model. In the field of medical image segmentation, some types of medical images are difficult to annotate, meaning that training samples are hard to obtain. With insufficient training samples, it is difficult to obtain a highly accurate image segmentation model, making it difficult to segment these types of medical images using the image segmentation model. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to overcome the difficulty in obtaining image samples for model training for certain types of medical images in the prior art. When the amount of image sample data is insufficient, the accuracy of the trained image segmentation model cannot be guaranteed, and therefore the image segmentation model cannot be used to segment this type of medical image. The present invention provides an image segmentation method, device, electronic device, and storage medium.

[0004] The present invention solves the above-mentioned technical problems through the following technical solution:

[0005] Firstly, an image segmentation method is provided, including:

[0006] Acquire first medical images;

[0007] The first medical image is converted into a second medical image of a target type, wherein the target type includes at least one of target phase, target contrast, target modality, and target resolution;

[0008] The second medical image is input into an image segmentation model to perform image segmentation processing on the region of interest in the second medical image; wherein, the image segmentation model is obtained by training a neural network with image samples of the target type, and the image samples are labeled with the identification information of the region of interest;

[0009] The segmentation result of the first medical image is determined based on the segmentation result of the second medical image.

[0010] Optionally, converting the first medical image into a second medical image of the target type includes:

[0011] The first medical image is input into an image conversion model to obtain the second medical image; wherein, the image conversion model is obtained by training a neural network with multiple sets of image samples, and each set of image samples includes at least one type of medical image sample of the scanned object and a medical image sample of the target type.

[0012] Optionally, the medical images included in each set of image samples are medical image samples that have undergone image registration processing.

[0013] Optionally, converting the first medical image into a second medical image of the target type includes:

[0014] If the target type includes the target resolution, the first medical image is sampled according to the target resolution to obtain a second medical image with the target resolution.

[0015] Optionally, before converting the first medical image into a second medical image of the target type, the method further includes:

[0016] Determine whether the type of the first medical image is the same as the target type;

[0017] If the determination result is negative, then the step of converting the first medical image into the second medical image is executed;

[0018] If the determination result is yes, then the first medical image is input into the image segmentation model to perform image segmentation processing on the first medical image.

[0019] Optionally, converting the first medical image into a second medical image of the target type includes:

[0020] If there are multiple image segmentation models corresponding to different target types, then the first medical image is converted into multiple second medical images corresponding to the different target types;

[0021] Determining the segmentation result of the first medical image based on the segmentation result of the second medical image includes:

[0022] The segmentation results of the multiple second medical images are weighted, and the segmentation result of the first medical image is determined based on the weighted processing result.

[0023] Optionally, it also includes:

[0024] When the target type includes the target resolution, the segmentation result of the second medical image is sampled according to the target resolution to obtain the segmentation result of the first medical image.

[0025] Secondly, an image segmentation apparatus is provided, comprising:

[0026] The acquisition module is used to acquire the first medical image;

[0027] A conversion module is used to convert the first medical image into a second medical image of a target type, wherein the target type includes at least one of target phase, target contrast, target modality, and target resolution;

[0028] An input module is used to input the second medical image into an image segmentation model to perform image segmentation processing on the region of interest in the second medical image; wherein, the image segmentation model is obtained by training a neural network with image samples of the target type, and the image samples include the identification information of the region of interest;

[0029] The determining module is used to determine the segmentation result of the first medical image based on the segmentation result of the second medical image.

[0030] Optionally, the conversion module is specifically used for:

[0031] The first medical image is input into an image conversion model to obtain the second medical image; wherein, the image conversion model is obtained by training a neural network with multiple sets of image samples, and each set of image samples includes at least one type of medical image sample of the scanned object and a medical image sample of the target type.

[0032] Optionally, the medical image samples included in each group of image samples are medical image samples that have undergone image registration processing.

[0033] Optionally, the conversion module is specifically used for:

[0034] If the target type includes the target resolution, the first medical image is sampled according to the target resolution to obtain a second medical image with the target resolution.

[0035] Optionally, it also includes:

[0036] The judgment module is used to determine whether the type of the first medical image is the same as the target type;

[0037] The judgment module is also used to call the conversion module if the judgment result is negative, and to call the input module if the judgment result is positive.

[0038] The input module is further configured to input the first medical image into the image segmentation model to perform image segmentation processing on the first medical image.

[0039] Optionally, the conversion module is specifically used for:

[0040] If there are multiple image segmentation models corresponding to different target types, then the first medical image is converted into multiple second medical images corresponding to the different target types;

[0041] The determining module is specifically used for:

[0042] The segmentation results of the multiple second medical images are weighted, and the segmentation result of the first medical image is determined based on the weighted processing result.

[0043] Optionally, it also includes:

[0044] A sampling module is used to sample the segmentation result of the second medical image according to the target resolution when the target type includes the target resolution, so as to obtain the segmentation result of the first medical image.

[0045] Thirdly, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the image segmentation method described in any of the preceding claims.

[0046] Fourthly, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the image segmentation method described in any of the preceding claims.

[0047] The positive and progressive effects of this invention are as follows: In the embodiments of this invention, the acquired medical image is converted into a target type of medical image, and the image segmentation processing of the acquired medical image is achieved by means of an image segmentation model corresponding to the target type. Therefore, it is not necessary to train multiple different types of image segmentation models, nor is it necessary to label the image samples used to train different types of image segmentation models, which can save a lot of manpower costs. Attached Figure Description

[0048] Figure 1 A flowchart of an image segmentation method provided as an exemplary embodiment of the present invention;

[0049] Figure 2 A flowchart illustrating another image segmentation method provided as an exemplary embodiment of the present invention;

[0050] Figure 3 A flowchart illustrating yet another image segmentation method provided as an exemplary embodiment of the present invention;

[0051] Figure 4 A schematic diagram of a module for an image segmentation apparatus provided in an exemplary embodiment of the present invention;

[0052] Figure 5This is a schematic diagram of the structure of an electronic device shown in an example embodiment of the present invention. Detailed Implementation

[0053] The present invention will be further illustrated by way of embodiments below, but the present invention is not limited to the scope of the embodiments described herein.

[0054] Figure 1 A flowchart of an image segmentation method provided as an exemplary embodiment of the present invention, the method comprising the following steps:

[0055] Step 101: Obtain the first medical image.

[0056] The first medical image is the medical image that requires image segmentation. The first medical image can be an image of any phase, any contrast, any modality, and any resolution.

[0057] In this context, "modality" refers to the type of medical imaging technology or imaging equipment. For example, a modality could be CT (Computed Tomography) modality, MRI (Magnetic Resonance Imaging) modality, PET (Positron Emission Tomography) modality, and so on. "Phase" refers to different imaging phases during the scanning process. For example, CT technology may include phases such as the plain scan phase, arterial phase, venous phase, and delayed phase; MRI technology may include phases such as T1, T2, DWI, and multi-phase enhanced images, with images of different phases obtained based on different scanning sequences.

[0058] For example, the first medical image could be a plain CT scan of the lungs. Distinguishing between pulmonary arteries and pulmonary veins in plain CT images is challenging, as is labeling them, making it difficult to obtain image samples for model training. Therefore, segmenting pulmonary arteries and pulmonary veins in plain CT images is a significant problem.

[0059] The first medical image can be a T2 phase image, that is, an image obtained based on a T2 sequence scan. Since it is difficult to clearly delineate the outline of the edema area in a T2 phase image, it is difficult to annotate it, which means it is difficult to obtain image samples for model training, making model training quite challenging. Therefore, segmenting the edema area in a T2 phase image is also a difficult problem.

[0060] Step 102: Convert the first medical image into a second medical image of the target type.

[0061] The target type of the second medical image includes at least one of target phase, target contrast, target modality, and target resolution. The target type of the second medical image is adapted to an existing or easily trained image segmentation model, which is used to perform image segmentation processing on the adapted target type of the second medical image.

[0062] Converting the first medical image into a second medical image allows for the segmentation of the second medical image (derived from the first image) using an existing image segmentation model designed for segmenting the second image. In other words, when an image segmentation model for segmenting the first medical image is unavailable, or when such a model is difficult to train, existing or easily trained image segmentation models are used to segment the first medical image.

[0063] In one embodiment, before converting the first medical image into a second medical image of the target type, it can be determined whether the type of the first medical image is the same as the target type. If the determination result is negative, it means that the existing or easily trained medical image segmentation model is not compatible with the type of the first medical image, and the image segmentation model cannot be used to directly segment the first medical image. In this case, step 102 is executed to convert the first medical image into a second medical image of the target type. If the determination result is positive, it means that the existing or easily trained image segmentation model is compatible with the image type of the first medical image, and the image segmentation model can be used to directly segment the first medical image. In this case, the first medical image is input into the compatible image segmentation model for image segmentation processing, without executing step 102.

[0064] Before converting the first medical image into a second medical image of the target type, it is determined whether there is an image segmentation model that matches the image type of the first medical image. If the determination result is yes, the existing image segmentation model is used to segment the first medical image, thus eliminating the need to convert the type of the first medical image, i.e., eliminating the need to perform the image conversion step, thereby reducing the amount of computation.

[0065] Before converting the first medical image into a second medical image of the target type, the type of the first medical image is used to determine whether an image segmentation model that is suitable for it is easy to train. If the determination result is yes, the model is trained for the first medical image to obtain an image segmentation model. The image segmentation model trained in a targeted manner is used to segment the first medical image, which can improve the accuracy of image segmentation.

[0066] The type of the first medical image can be determined by a trained image type recognition module; or it can be determined by reading the image information of the first medical image, such as tag information in a DICOM (Digital Imaging and Communications in Medicine) file. This tag information includes at least one of the following scanning parameters: phase-related scanning parameters, resolution-related scanning parameters, contrast-related scanning parameters, modality-related scanning parameters, etc.

[0067] In one embodiment, an image conversion model is used to convert a first medical image into a second medical image. Specifically, the first medical image is input into the image conversion model, and the second medical image is obtained based on the output of the image conversion model.

[0068] The image conversion model is trained on a neural network using multiple sets of image samples. Each set of image samples includes at least one type of medical image sample of the scanned object and a target type of medical image sample. The target type of medical image sample serves as the gold standard during model training. During training, the at least one type of medical image sample is input into the neural network, the output of the neural network is obtained, and the network parameters are adjusted based on the loss error between the output and the target type of medical image sample. This process is repeated until the training stopping condition is met, and the trained neural network is then identified as the image conversion model.

[0069] It's important to note that to determine the type of medical image the image conversion model will output, the corresponding medical image samples from each set of image samples are used as the gold standard during model training. For example, if the goal is for the image conversion model to output a medical image of the venous phase, both plain CT scan images and venous phase images are used as image samples, with the venous phase images serving as the gold standard during training. The resulting image conversion model can then convert an input plain CT scan image (the first medical image) into a venous phase image (the second medical image). Therefore, with a sufficiently rich training sample, the image conversion model can convert any type of first medical image into a second medical image, demonstrating high versatility and ease of operation. The network architecture of the image conversion model can be, but is not limited to, U-Net, VB-Net, ResUNet, GAN, or Cycle-GAN.

[0070] Each medical image sample in each group of images is an image obtained from scanning the object at different times. During the scanning process, due to muscle contraction, respiratory movements, etc. of the scanned object, there will be positional deviations in the content displayed in each medical image. Therefore, in order to improve the accuracy of the image conversion model, before using the image samples to train the neural network, it is necessary to perform image registration processing on the medical image samples in each group of images. The neural network is then trained using the medical image samples after image registration processing to obtain the image conversion model.

[0071] In one embodiment, if the target type includes a target resolution, i.e., the first medical image needs to be converted into a second medical image at the target resolution, in addition to obtaining the second medical image at the target resolution based on an image conversion model, the first medical image can also be sampled according to the target resolution to obtain the second medical image at the target resolution. Obtaining the second medical image based on sampling eliminates the need for model training and reduces computational cost.

[0072] Understandably, an image segmentation model can be a single model or multiple models that segment different types of medical images. If multiple image segmentation models exist corresponding to different target types, the first medical image is converted into multiple second medical images corresponding to different target types, resulting in multiple second medical images. Subsequently, image segmentation of the first medical image is achieved based on these multiple second medical images.

[0073] Step 103: Input the second medical image into the image segmentation model to perform image segmentation processing on the region of interest in the second medical image.

[0074] This image segmentation model is trained on a neural network using image samples of the target type, including region-of-interest (ROI) labeling information. This labeling information serves as the gold standard during model training. During training, image samples of the target type are input into the neural network, and the network output is obtained. The network parameters are adjusted based on the loss error between the output and the labeling information of the image samples. This process is repeated until the training stopping condition is met, at which point the trained neural network is identified as the image segmentation model.

[0075] The image segmentation model is either an existing one or trained on image samples that are relatively easy to obtain (easy and accurate labeling). When image samples are readily available, a rich and sufficient amount of image samples can be obtained, leading to a more convergent and higher segmentation accuracy image segmentation model. The network architecture of the image segmentation model can, but is not limited to, DenseNet, FPN, or V-Net networks.

[0076] Understandably, training a neural network with image samples of the target type, that is, training the neural network specifically based on image samples of the target type, results in higher image segmentation accuracy when using an image segmentation model trained specifically for the target type of medical images compared to a segmentation model that is relatively balanced across all types of medical images.

[0077] If there are multiple image segmentation models corresponding to different target types, the first medical image is converted into multiple second medical images corresponding to different target types, and the multiple second medical images are respectively input into the image segmentation model that is compatible with their types, so as to perform image segmentation processing on the regions of interest in the multiple second medical images respectively, and obtain multiple image segmentation results. In step 104 below, the segmentation result of the first medical image is determined based on the multiple image segmentation results.

[0078] For example, assuming the first medical image is a plain scan image, and there are two existing image segmentation models for target types, namely image segmentation model A for segmenting arterial images and image segmentation model B for segmenting venous images, then by inputting the first medical image into image segmentation model A and image segmentation model B respectively, two segmentation results are obtained, namely the segmentation result output by image segmentation model A and the segmentation result output by image segmentation model B, resulting in multiple image segmentation results.

[0079] Step 104: Determine the segmentation result of the first medical image based on the segmentation result of the second medical image.

[0080] The segmentation results of medical images include the location and shape information of the region of interest in the medical image.

[0081] Since the second medical image in the input image segmentation model is obtained by transforming the first medical image, the shape and position of the region of interest in the second medical image correspond to its shape and position in the first medical image. Therefore, the segmentation result of the first medical image can be determined based on the segmentation result of the second medical image.

[0082] If step 103 yields an image segmentation result, step 104 will directly determine this image segmentation result as the segmentation result of the first medical image.

[0083] If multiple image segmentation results are obtained in step 103, the multiple image segmentation results are fused in step 104 to obtain the segmentation result of the first medical image.

[0084] In one embodiment, fusing multiple image segmentation results can be achieved by weighting the segmentation results of each second medical image, and then using the weighted result as the segmentation result of the first medical image. The segmentation result of the first medical image is obtained by combining the segmentation results of multiple second medical images, resulting in a more accurate segmentation result.

[0085] In one embodiment, if the target type includes a target resolution, it is also necessary to sample the segmentation result of the second medical image according to the target resolution to obtain the segmentation result of the target resolution, and the segmentation result of the first medical image is determined based on the segmentation result obtained by sampling.

[0086] In this embodiment of the invention, the acquired medical image is converted into a target type of medical image, and image segmentation processing of the acquired medical image is achieved by using an image segmentation model corresponding to the target type. This eliminates the need to train multiple image segmentation models of different types, and eliminates the need to label image samples used to train different types of image segmentation models, especially eliminating the need to label difficult-to-label image samples, thus saving significant manpower costs. The image segmentation method provided by this embodiment of the invention can be extended to various scenarios with different types of image segmentation requirements.

[0087] exist Figure 1 On this basis, Figure 2 A flowchart of another image segmentation method is shown, which includes the following steps:

[0088] Step 201: Obtain the first medical image.

[0089] Step 202: Identify the type of the first medical image.

[0090] The type of the first medical image can be determined by a trained image type recognition module; or it can be determined by reading the image information of the first medical image, such as tag information in a DICOM (Digital Imaging and Communications in Medicine) file. This tag information includes at least one of the following scanning parameters: phase-related scanning parameters, resolution-related scanning parameters, contrast-related scanning parameters, modality-related scanning parameters, etc.

[0091] Step 203: Determine whether the type of the first medical image is the same as the target type.

[0092] If the judgment result is negative, that is, the type of the first medical image is different from the target type, then proceed to step 204; if the judgment result is positive, that is, the type of the first medical image is the same as the target type, then proceed to step 206.

[0093] Step 204: Convert the first medical image into a second medical image of the target type.

[0094] Step 205: Input the second medical image into the existing image segmentation model to perform image segmentation processing on the second medical image.

[0095] In step 205, the image segmentation model is used to perform image segmentation processing on the second medical image.

[0096] The implementation of steps 204 and 205 is similar to that of steps 102 and 103. The specific implementation process of steps 204 and 205 will not be described here.

[0097] Step 206: Input the first medical image into the existing image segmentation model to perform image segmentation processing on the first medical image.

[0098] In step 206, the image segmentation model is used to perform image segmentation processing on the first medical image.

[0099] Step 207: Determine the segmentation result of the first medical image.

[0100] In step 207, the output of the image segmentation model is determined as the segmentation result of the first medical image. If the type of the first medical image is the same as the target type, the output of the image segmentation model is the segmentation result of the first medical image. If the type of the first medical image is different from the target type, the output of the image segmentation model is the segmentation result of the second medical image. Since the second medical image is obtained by transforming the first medical image, the shape and position of the region of interest in the second medical image correspond to its shape and position in the first medical image. Therefore, the segmentation result of the second medical image can represent the segmentation result of the first medical image.

[0101] Figure 3 This is a flowchart illustrating another image segmentation method according to an embodiment of the present invention. The image segmentation method in this embodiment is similar to... Figure 2 The image segmentation methods shown are basically the same, the difference being that in this embodiment, multiple image segmentation results are obtained, and these results need to be fused. See also Figure 3 The method includes the following steps:

[0102] Step 301: Obtain the first medical image.

[0103] Step 302: Identify the type of the first medical image.

[0104] The type of the first medical image can be determined by a trained image type recognition module; or it can be determined by reading the image information of the first medical image, such as tag information in a DICOM (Digital Imaging and Communications in Medicine) file. This tag information includes at least one of the following scanning parameters: phase-related scanning parameters, resolution-related scanning parameters, contrast-related scanning parameters, modality-related scanning parameters, etc.

[0105] Step 303: Determine whether the type of the first medical image is the same as the target type.

[0106] If the judgment result is negative, that is, the type of the first medical image is different from the target type, then proceed to step 304; if the judgment result is positive, that is, the type of the first medical image is the same as the target type, then proceed to step 307.

[0107] Step 304: Convert the first medical image into a second medical image of the target type.

[0108] Step 305: Input the second medical image into the existing image segmentation model to perform image segmentation processing on the second medical image.

[0109] In step 305, the image segmentation model is used to perform image segmentation processing on the second medical image.

[0110] The implementation of steps 304 and 305 is similar to that of steps 102 and 103. The specific implementation process of steps 304 and 305 will not be described here.

[0111] If multiple segmentation results are obtained in step 305, then step 306 is executed; otherwise, step 308 is executed.

[0112] Step 306: Merge the segmentation results of multiple images.

[0113] Step 307: Input the first medical image into the existing image segmentation model to perform image segmentation processing on the first medical image.

[0114] In step 307, the image segmentation model is used to perform image segmentation processing on the first medical image.

[0115] Step 308: Determine the segmentation result of the first medical image.

[0116] If the first medical image type is the same as the target type, in step 308, the output result of the image segmentation model is determined as the segmentation result of the first medical image.

[0117] If the first medical image type is different from the target type, and there are multiple image segmentation models for image segmentation processing of the second medical image, in step 308, the fusion result obtained by fusing the output results of the multiple image segmentation models is determined as the segmentation result of the first medical image.

[0118] In another embodiment, an image segmentation model trained based on more readily available image samples is used instead. Figure 2 , Figure 3 The existing image segmentation model is shown in the figure.

[0119] Corresponding to the aforementioned image segmentation method embodiments, the present invention also provides embodiments of an image segmentation apparatus.

[0120] Figure 4 A schematic diagram of an image segmentation apparatus provided as an exemplary embodiment of the present invention, the image segmentation apparatus comprising:

[0121] Acquisition module 41 is used to acquire the first medical image;

[0122] The conversion module 42 is used to convert the first medical image into a second medical image of a target type, wherein the target type includes at least one of target phase, target contrast, target modality, and target resolution;

[0123] The input module 43 is used to input the second medical image into the image segmentation model to perform image segmentation processing on the region of interest in the second medical image; wherein, the image segmentation model is obtained by training a neural network with image samples of the target type, and the image samples include the identification information of the region of interest;

[0124] The determining module 44 is used to determine the segmentation result of the first medical image based on the segmentation result of the second medical image.

[0125] Optionally, the conversion module is specifically used for:

[0126] The first medical image is input into an image conversion model to obtain the second medical image; wherein, the image conversion model is obtained by training a neural network with multiple sets of image samples, and each set of image samples includes at least one type of medical image sample of the scanned object and a medical image sample of the target type.

[0127] Optionally, the medical images included in each set of image samples are medical images that have undergone image registration processing.

[0128] Optionally, the conversion module is specifically used for:

[0129] If the target type includes the target resolution, the first medical image is resampled according to the target resolution to obtain a second medical image with the target resolution.

[0130] Optionally, it also includes:

[0131] The judgment module is used to determine whether the type of the first medical image is the same as the target type;

[0132] The judgment module is also used to call the conversion module if the judgment result is negative, and to call the input module if the judgment result is positive.

[0133] The input module is further configured to input the first medical image into the image segmentation model to perform image segmentation processing on the first medical image.

[0134] Optionally, the conversion module is specifically used for:

[0135] If there are multiple image segmentation models corresponding to different target types, then the first medical image is converted into multiple second medical images corresponding to the different target types;

[0136] The determining module is specifically used for:

[0137] The segmentation results of each second medical image are weighted and the segmentation result of the first medical image is determined based on the weighted result.

[0138] Optionally, it also includes:

[0139] A resampling module is used to resample the segmentation result of the second medical image according to the target resolution when the target type includes the target resolution; the segmentation result of the first medical image is determined based on the resampled segmentation result.

[0140] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of the present invention according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0141] Figure 5This is a schematic diagram of the structure of an electronic device according to an example embodiment of the present invention, showing a block diagram of an exemplary electronic device 50 suitable for implementing embodiments of the present invention. Figure 5 The electronic device 50 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of the present invention.

[0142] like Figure 5 As shown, the electronic device 50 can be manifested in the form of a general-purpose computing device, such as a server device. The components of the electronic device 50 may include, but are not limited to: at least one processor 51, at least one memory 52, and a bus 53 connecting different system components (including memory 52 and processor 51).

[0143] Bus 53 includes a data bus, an address bus, and a control bus.

[0144] The memory 52 may include volatile memory, such as random access memory (RAM) 521 and / or cache memory 522, and may further include read-only memory (ROM) 523.

[0145] The memory 52 may also include a program tool 525 (or utility) having a set (at least one) program module 524, such program module 524 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0146] The processor 51 performs various functional applications and data processing, such as the methods provided in any of the above embodiments, by running computer programs stored in the memory 52.

[0147] Electronic device 50 can also communicate with one or more external devices 54 (e.g., keyboard, pointing device, etc.). This communication can be performed via input / output (I / O) interface 55. Furthermore, the model-generated electronic device 50 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 56. As shown, network adapter 56 communicates with other modules of the model-generated electronic device 50 via bus 53. It should be understood that, although not shown in the figure, other hardware and / or software modules can be used in conjunction with the model-generated electronic device 50, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (disk array) systems, tape drives, and data backup storage systems.

[0148] It should be noted that although several units / modules or sub-units / modules of the electronic device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided and embodied by multiple units / modules.

[0149] It should be noted that although several units / modules or sub-units / modules of the electronic device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided and embodied by multiple units / modules.

[0150] This invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method provided in any of the above embodiments.

[0151] The readable storage medium may be more specifically adopted, including but not limited to: portable disk, hard disk, random access memory, read-only memory, erasable programmable read-only memory, optical storage device, magnetic storage device, or any suitable combination thereof.

[0152] In a possible implementation, the present invention can also be implemented as a program product comprising program code, wherein when the program product is run on a terminal device, the program code is used to cause the terminal device to execute the method implementing any of the above embodiments.

[0153] The program code for executing the present invention can be written in any combination of one or more programming languages. The program code can be executed entirely on the user device, partially on the user device, as a standalone software package, partially on the user device and partially on a remote device, or entirely on a remote device.

[0154] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these are merely illustrative examples, and the scope of protection of the present invention is defined by the appended claims. Those skilled in the art can make various changes or modifications to these embodiments without departing from the principles and essence of the present invention, but all such changes and modifications fall within the scope of protection of the present invention.

Claims

1. An image segmentation method, characterized in that, include: Acquire first medical images; The first medical image is converted into a second medical image of a target type, wherein the target type includes a target phase. The second medical image is input into an image segmentation model to perform image segmentation processing on the region of interest in the second medical image; wherein, the image segmentation model is obtained by training a neural network with image samples of the target type, and the image samples include the identification information of the region of interest; The segmentation result of the first medical image is determined based on the segmentation result of the second medical image; Converting the first medical image into a second medical image of the target type includes: If there are multiple image segmentation models corresponding to different target types, then the first medical image is converted into multiple second medical images corresponding to the different target types; Determining the segmentation result of the first medical image based on the segmentation result of the second medical image includes: The segmentation results of the plurality of second medical images are weighted, and the segmentation result of the first medical image is determined based on the weighting result. Converting the first medical image into a second medical image of the target type includes: The first medical image is input into an image conversion model to obtain the second medical image; wherein the image conversion model is obtained by training a neural network with multiple sets of image samples, and each set of image samples includes at least one type of medical image sample of the scanned object and a medical image sample of the target type.

2. The image segmentation method according to claim 1, characterized in that, Each set of image samples contains medical image samples that have undergone image registration processing.

3. The image segmentation method according to claim 1, characterized in that, Converting the first medical image into a second medical image of the target type includes: If the target type includes a target resolution, the first medical image is sampled according to the target resolution to obtain a second medical image with the target resolution.

4. The image segmentation method according to claim 1, characterized in that, Before converting the first medical image into a second medical image of the target type, the process also includes: Determine whether the type of the first medical image is the same as the target type; If the determination result is negative, then the step of converting the first medical image into the second medical image is executed; If the determination result is yes, then the first medical image is input into the image segmentation model to perform image segmentation processing on the first medical image.

5. The image segmentation method according to any one of claims 1-4, characterized in that, Also includes: When the target type includes a target resolution, the segmentation result of the second medical image is sampled according to the target resolution to obtain the segmentation result of the first medical image.

6. An image segmentation apparatus, characterized in that, include: The acquisition module is used to acquire the first medical image; A conversion module is used to convert the first medical image into a second medical image of a target type, wherein the target type includes a target phase. An input module is used to input the second medical image into an image segmentation model to perform image segmentation processing on the region of interest in the second medical image; wherein, the image segmentation model is obtained by training a neural network with image samples of the target type, and the image samples are labeled with the identification information of the region of interest; A determining module is used to determine the segmentation result of the first medical image based on the segmentation result of the second medical image; The conversion module is specifically used to: if there are multiple image segmentation models corresponding to different target types, convert the first medical image into multiple second medical images corresponding to the different target types; The determining module is specifically used to: perform weighted processing on the segmentation results of the plurality of second medical images, and determine the segmentation result of the first medical image based on the weighted processing result; The conversion module is specifically used for: The first medical image is input into an image conversion model to obtain the second medical image; wherein the image conversion model is obtained by training a neural network with multiple sets of image samples, and each set of image samples includes at least one type of medical image sample of the scanned object and a medical image sample of the target type.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the image segmentation method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the image segmentation method according to any one of claims 1 to 5.

Images