Focus segmentation method and device, medium and product

By resampling and pre-training model processing of prostate cancer images, the problem of low accuracy of lesion segmentation in the prior art is solved, and high accuracy segmentation of multiple lesions is achieved, ensuring the detection of small lesions.

CN119068197APending Publication Date: 2024-12-03SHANGHAI HAIWEI MEDICAL TECH CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202411206085.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-30
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

The prior art has low accuracy in the lesion segmentation results in prostate cancer images, making it difficult to accurately segment multiple lesions of different sizes.

Method used

By resampling the first image, the second image is obtained, and the voxel size is smaller than the original image. The pre-trained model is input to obtain the lesion segmentation result, and the results are resampled to match the voxel size of the original image, and the segmentation result is combined to improve the accuracy.

Benefits of technology

It improves the accuracy of lesion segmentation, ensures that small lesions are not missed, and enhances the ability to segment multiple lesions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119068197A_ABST
    Figure CN119068197A_ABST
Patent Text Reader

Abstract

The invention discloses a focus segmentation method and device, a medium and a product. The first image is resampled to obtain a second image, the voxel size in the second image is smaller than that in the first image, the first image comprises at least one first focus and at least one second focus, and the volume of the second focus is smaller than that of the first focus; respectively inputting the first image and the second image into a pre-training first model and a pre-training second model to obtain a first focus segmentation result and a second focus segmentation result; re-sampling the second focus segmentation result to obtain a third focus segmentation result, wherein the voxel size in the third focus segmentation result is the same as the voxel size in the first image; and determining a target lesion segmentation result according to the first lesion segmentation result and the third lesion segmentation result. According to the technical scheme provided by the embodiment of the invention, the focus segmentation accuracy can be improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing, and in particular to a lesion segmentation method, equipment, medium and product. Background Art

[0002] Prostate cancer images usually include multiple lesions of different sizes. The existing technology generally uses threshold segmentation or model segmentation to segment the lesions, but it is difficult to accurately segment each lesion.

[0003] Therefore, it is necessary to provide a lesion segmentation method to improve the accuracy of lesion segmentation. Summary of the invention

[0004] The present invention provides a lesion segmentation method, device, medium and product to solve the problem of low accuracy of lesion segmentation results in the prior art lesion segmentation methods.

[0005] According to one aspect of the present invention, a lesion segmentation method is provided, comprising:

[0006] Resampling the first image to obtain a second image, wherein a voxel size in the second image is smaller than a voxel size in the first image, the first image includes at least one first lesion and at least one second lesion, and a volume of the second lesion is smaller than a volume of the first lesion;

[0007] Inputting the first image and the second image into a pre-trained first model and a pre-trained second model respectively to obtain a first lesion segmentation result and a second lesion segmentation result;

[0008] resampling the second lesion segmentation result to obtain a third lesion segmentation result, wherein the voxel size in the third lesion segmentation result is the same as the voxel size in the first image;

[0009] A target lesion segmentation result is determined according to the first lesion segmentation result and the third lesion segmentation result.

[0010] According to another aspect of the present invention, there is provided a lesion segmentation device, comprising:

[0011] A first resampling module is used to resample the first image to obtain a second image, wherein the voxel size in the second image is smaller than the voxel size in the first image, the first image includes at least one first lesion and at least one second lesion, and the volume of the second lesion is smaller than the volume of the first lesion;

[0012] A segmentation module, inputting the first image and the second image into a pre-trained first model and a pre-trained second model respectively, to obtain a first lesion segmentation result and a second lesion segmentation result;

[0013] A second resampling module is used to resample the second lesion segmentation result to obtain a third lesion segmentation result, wherein the voxel size in the third lesion segmentation result is the same as the voxel size in the first image;

[0014] A result module is used to determine a target lesion segmentation result according to the first lesion segmentation result and the third lesion segmentation result.

[0015] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0016] at least one processor; and

[0017] a memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the lesion segmentation method described in any embodiment of the present invention.

[0019] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the lesion segmentation method described in any embodiment of the present invention when executed.

[0020] According to another aspect of the present invention, a computer program product is provided. The computer program product comprises a computer program. When the computer program is executed by a processor, the computer program implements the lesion segmentation method described in any embodiment.

[0021] The technical solution provided by the embodiment of the present invention obtains a second image with higher spatial resolution by reducing the voxel size of the first image; the first image and the second image are processed respectively by a pre-trained first model and a pre-trained second model to obtain a first lesion segmentation result and a second lesion segmentation result. Since the voxel size corresponding to the second image is smaller than the voxel size corresponding to the first image, the second lesion segmentation result includes some small lesion segmentation results that do not exist in the first lesion segmentation result. Therefore, the second lesion segmentation result is converted from the second accuracy to the first accuracy to obtain a third lesion segmentation result. The target lesion segmentation result corresponding to the union of the first lesion segmentation result and the third lesion segmentation result has high accuracy and no small lesions will be missed.

[0022] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 is a flow chart of a lesion segmentation method provided according to an embodiment of the present invention;

[0025] Figure 2 is another flow chart of a lesion segmentation method provided according to an embodiment of the present invention;

[0026] Figure 3 is a flowchart of a model training method provided according to an embodiment of the present invention;

[0027] Figure 4 is a schematic structural diagram of a lesion segmentation device provided according to an embodiment of the present invention;

[0028] Figure 5 is another structural schematic diagram of a lesion segmentation device provided according to an embodiment of the present invention;

[0029] Figure 6 is another structural schematic diagram of a lesion segmentation device provided according to an embodiment of the present invention;

[0030] Figure 7 It is a schematic diagram of the structure of an electronic device for implementing the lesion segmentation method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0032] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0033] Figure 1 This is a flow chart of a lesion segmentation method provided by an embodiment of the present invention. This embodiment is applicable to the case of automatically completing lesion segmentation of medical images. The method can be performed by a lesion segmentation device. The lesion segmentation device can be implemented in the form of hardware and / or software. The lesion segmentation device can be configured in an electronic device. Figure 1 As shown, the method includes:

[0034] S110. Resample the first image to obtain a second image, wherein a voxel size in the second image is smaller than a voxel size in the first image, the first image includes at least one first lesion and at least one second lesion, and a volume of the second lesion is smaller than a volume of the first lesion.

[0035] The first image is a medical image, such as a CT image, a PET image, a PET-CT image, or an MRI image.

[0036] In one embodiment, the initial image is a PET (Positron Emission Tomography) image used for clinical diagnosis, such as a prostate PET image used for clinical diagnosis. The format of the initial image can be selected as a nii file.

[0037] The smaller the voxel size, the smaller its volume, and the higher the spatial resolution of the corresponding image. The first image is resampled to obtain a second image, and the voxel size in the second image is smaller than the voxel size in the first image, so that the spatial resolution of the second image is greater than the spatial resolution of the first image. The first image includes at least one first lesion and at least one second lesion. The first lesion is larger than the second lesion.

[0038] In one embodiment, in order to improve the accuracy of the first lesion segmentation result, the first image is resampled to update the first image, and the voxel size in the updated first image is larger than the voxel size in the second image, but smaller than the voxel size in the first image. In other words, the spatial resolution of the updated first image is smaller than the spatial resolution of the second image, but larger than the spatial resolution before the update.

[0039] S120, inputting the first image and the second image into a pre-trained first model and a pre-trained second model respectively to obtain a first lesion segmentation result and a second lesion segmentation result.

[0040] Since the first lesion is larger than the second lesion, the first lesion segmentation result can also be called a large lesion segmentation result, and the second lesion segmentation result can also be called a small lesion segmentation result.

[0041] The pre-trained first model is configured to determine a first lesion segmentation result of the first image. The pre-trained second model is configured to determine a second lesion segmentation result of the second image. Therefore, the pre-trained first model is used to segment large lesions, and the pre-trained second model is used to segment small lesions.

[0042] The first model and the second model can adopt the same model structure or different model structures. In one embodiment, the first model and the second model adopt the same model structure, including a first dual convolution layer, a second dual convolution layer, a first downsampling module, a second downsampling module, a third downsampling module, a fourth downsampling module, a first upsampling module, a second upsampling module, a third upsampling module and a fourth upsampling module connected in sequence, and the output end of the fourth downsampling module is also connected to the input end of the first upsampling module, the output end of the third downsampling module is also connected to the input end of the second upsampling module, the output end of the second downsampling module is also connected to the input end of the third upsampling module, and the output end of the first downsampling module is also connected to the output end of the fourth upsampling module. Among them, the dual convolution layer includes a three-dimensional convolution layer, a normalization layer, an activation function layer, a three-dimensional convolution layer, a normalization layer, and an activation function layer connected in sequence. The downsampling module includes a dual convolution layer and a maximum pooling downsampling layer connected in sequence. The upsampling module includes a transposed convolution layer and a dual convolution layer connected in sequence.

[0043] S130 , resampling the second lesion segmentation result to obtain a third lesion segmentation result, wherein the voxel size in the third lesion segmentation result is the same as the voxel size in the first image.

[0044] The first lesion segmentation result can be regarded as the probability that each voxel in the first image is the first lesion; the second lesion segmentation result can be regarded as the probability that each voxel in the second image is the second lesion. The probability threshold is read, and the voxels in the first lesion segmentation result that are greater than or equal to the probability threshold are regarded as valid voxels, otherwise, they are removed from the first lesion segmentation result as invalid voxels to update the first lesion segmentation result; similarly, the voxels in the second lesion segmentation result that are greater than or equal to the probability threshold are regarded as valid voxels, otherwise, they are removed from the second lesion segmentation result as invalid voxels to update the second lesion segmentation result.

[0045] The voxel size corresponding to the first lesion segmentation result is the same as the voxel size corresponding to the first image, and the voxel size corresponding to the second lesion segmentation result is the same as the voxel size corresponding to the second image. Therefore, the voxel size corresponding to the current second lesion segmentation result is converted to the voxel size corresponding to the first image to obtain the third lesion segmentation result.

[0046] It can be understood that if the voxel size of the current first image is the voxel size after resampling, the voxel size of the current first image is converted to the voxel size before resampling.

[0047] In one embodiment, after the third lesion segmentation result is determined, lesions whose lesion volumes are greater than a set volume threshold are deleted from the third lesion segmentation result to obtain an updated third lesion segmentation result. This embodiment enables the updated third lesion distribution result to include only small lesions whose volumes are less than or equal to the set volume threshold.

[0048] In one embodiment, lesions whose volumes are greater than a set volume threshold may be deleted from the second lesion segmentation result to obtain an updated second lesion segmentation result. The updated second lesion segmentation result is then resampled to obtain a fourth lesion segmentation result. It is understandable that the fourth lesion segmentation result is the same as the updated third lesion segmentation result. This embodiment enables the fourth lesion segmentation result to include only small lesions whose volumes are less than or equal to the set volume threshold.

[0049] S140: Determine a target lesion segmentation result according to the first lesion segmentation result and the third lesion segmentation result.

[0050] The first image is an image that can be used for clinical positioning, and the base plate image corresponding to the first image is determined.

[0051] In one embodiment, a union of the first lesion segmentation result and the latest third lesion segmentation result is determined, and the union is added to the base plate image to obtain the target lesion segmentation result of the first image.

[0052] In yet another embodiment, the first lesion segmentation result and the third lesion segmentation result are both added to the base image to obtain the target lesion segmentation result of the first image.

[0053] It can be understood that the target lesion segmentation result includes the segmentation result of a large lesion and the segmentation result of a small lesion.

[0054] In one embodiment, the training of the first model is completed by the following method:

[0055] Step a1: Acquire multiple first images for model training.

[0056] Step a2, assigning zero to the non-lesion area in each first image as a label of each first image;

[0057] Step a3: input each first image and the label of each first image into the first model to obtain a first intermediate segmentation result, and adjust the model parameters according to the comparison result between the first intermediate segmentation result and the corresponding label until the model parameters of the first model meet the set parameter conditions, thereby obtaining a pre-trained first model.

[0058] In one embodiment, the first image is a part of an image that can be used for clinical positioning. At this time, an initial image is acquired, and the initial image is cropped based on SUV (standard uptake value) information to obtain a first image. In each initial image, the maximum and minimum values ​​of the SUV information in each coordinate direction are determined, and the initial image is cropped based on the maximum and minimum values ​​to obtain the first image, and the maximum and minimum values ​​are divided into the first image. This embodiment aims to reduce the amount of data input to the model under the premise that the number of voxels carrying SUV information in the first image is equal to the number of voxels carrying SUV information in the initial image.

[0059] The technical solution provided by the embodiment of the present invention obtains a second image with higher spatial resolution by reducing the voxel size of the first image; the first image and the second image are processed respectively by a pre-trained first model and a pre-trained second model to obtain a first lesion segmentation result and a second lesion segmentation result. Since the voxel size corresponding to the second image is smaller than the voxel size corresponding to the first image, the second lesion segmentation result includes some small lesion segmentation results that do not exist in the first lesion segmentation result. Therefore, the second lesion segmentation result is converted from the second accuracy to the first accuracy to obtain a third lesion segmentation result. The target lesion segmentation result corresponding to the union of the first lesion segmentation result and the third lesion segmentation result has high accuracy and no small lesions will be missed.

[0060] Figure 2FIG. 1 is a flow chart of a lesion segmentation method provided in an embodiment of the present invention. This embodiment adds an image cropping step based on the above-mentioned embodiment. Figure 2 As shown, the method includes:

[0061] S200 , cropping the initial image based on the SUV information to obtain a first image, wherein the number of voxels carrying the SUV information in the first image is equal to the number of voxels carrying the SUV information in the initial image.

[0062] An initial image is acquired, where the initial image can be used for clinical positioning.

[0063] After the initial image is acquired, it is determined whether the SUV information is included; if not, the SUV information is acquired; if the SUV information is included, the maximum and minimum values ​​of the SUV information distribution range in each coordinate direction are determined, and the initial image is cropped according to the maximum and minimum values ​​of the SUV information distribution range in each coordinate direction to obtain the first image, and the maximum and minimum values ​​of the SUV information distribution range in each coordinate direction are located in the first image. In this way, the first image includes all SUV information distribution information in the initial image.

[0064] S210. Resample the first image to obtain a second image, wherein a voxel size in the second image is smaller than a voxel size in the first image, the first image includes at least one first lesion and at least one second lesion, and a volume of the second lesion is smaller than a volume of the first lesion.

[0065] In one embodiment, after the first image and the second image are determined, SUV information normalization is performed on the first image and the second image respectively to obtain an updated first image and an updated second image. This embodiment improves the accuracy of the lesion segmentation results determined by each model by normalizing the SUV information.

[0066] Specifically, the normalization method of the first image and the second image is the same, and the normalization of the first image is taken as an example. The mean and variance of the SUV information of each voxel in the first image are determined, and the first image is normalized based on the mean and variance of the SUV information of each voxel to obtain a normalized first image.

[0067] S220, inputting the first image and the second image into a pre-trained first model and a pre-trained second model respectively to obtain a first lesion segmentation result and a second lesion segmentation result.

[0068] S230 , resampling the second lesion segmentation result to obtain a third lesion segmentation result, wherein the voxel size in the third lesion segmentation result is the same as the voxel size in the first image.

[0069] S2401. Determine two base images corresponding to the initial image, wherein the voxel value of each voxel in the base images is zero.

[0070] Since the initial image carries coordinate information for positioning, the lesion segmentation result needs to be filled back into the initial image. To this end, the present embodiment determines two base images corresponding to the initial image, and the voxel value of each voxel in the base image is zero.

[0071] S2402. According to the position of the first image in the initial image, add the first lesion segmentation result to one base plate image to update the first lesion segmentation result, and add the third lesion segmentation result to another base plate image to update the third lesion segmentation result.

[0072] According to the position of the first image in the initial image, the first lesion segmentation result is added to one base plate image to update the first lesion segmentation result, and the third lesion segmentation result is added to another base plate image to update the third lesion segmentation result. In this way, the spatial position of each lesion in the updated first lesion segmentation result is its position in the initial image. The spatial position of each lesion in the updated second lesion segmentation result is its position in the initial image.

[0073] In one embodiment, lesions whose volumes are greater than a set volume threshold are deleted from the updated third lesion segmentation result to obtain a fourth lesion segmentation result; and the updated first lesion segmentation result and the fourth lesion segmentation result are used as target segmentation results.

[0074] Since the first image includes both the first lesion and the second lesion, the second lesion segmentation result includes the first lesion (large lesion) and the second lesion (small lesion). Since the pre-trained second lesion is trained to segment the second lesion, the accuracy of the second lesion is higher. Therefore, the lesions with a volume greater than the set volume threshold in the updated third lesion segmentation result are deleted to obtain the fourth lesion segmentation result. It can be understood that the large lesion in the updated third lesion segmentation result is deleted to obtain the fourth lesion segmentation result, and the fourth lesion segmentation result only includes the small lesion in the updated third lesion segmentation result.

[0075] S2403: Taking the updated first lesion segmentation result and the third lesion segmentation result as a target segmentation result.

[0076] In one embodiment, the first lesion segmentation result is superimposed on the third lesion segmentation result to obtain a target segmentation result. This embodiment can quickly and accurately determine the target lesion segmentation result.

[0077] S2403. Display the target lesion segmentation result.

[0078] After the target lesion segmentation result is determined, the target lesion segmentation result is sent to the front end so that the front end displays the target lesion segmentation result.

[0079] In an embodiment of the present invention, an initial image is cropped based on SUV information to obtain a first image, and a target lesion segmentation result determined based on the first image is filled back into a base image corresponding to the initial image. This not only reduces the data processing amount of the model and improves the data processing speed of the model, but also ensures the accuracy of the target lesion segmentation result.

[0080] Figure 3 Flow chart of the model training method provided by the embodiment of the present invention. Figure 3 As shown, the method includes:

[0081] S310, obtaining a training sample set, the training sample set including at least two training samples, the training samples including a label and a lesion whose volume is smaller than a set volume threshold, the label being that the voxel value of the voxel corresponding to the lesion whose volume is smaller than the set volume threshold is 1, and the voxel value of other voxels is zero.

[0082] After obtaining multiple first images, first determine the label of each first image. Specifically, select a first image randomly from the multiple first images, and then assign zero to the lesions in the first image whose volume is greater than the set volume threshold, and then determine the threshold of the remaining lesions based on the Kmeans algorithm, and use the threshold as the set volume threshold. In the first image, the lesions whose volume is less than or equal to the set volume threshold are assigned 1, and the rest are assigned zero. Then, the lesions in the remaining first images whose volume is less than or equal to the set threshold are assigned 1, and the rest are assigned zero to obtain the label of the corresponding first image.

[0083] After the labels of each first image are determined, the lesions in the initial first image whose volumes are larger than the set volume threshold are assigned zero and used as training samples.

[0084] S320 , resample each training sample to update each training sample, and the voxel size of each updated training sample is the same as the voxel size of the first image.

[0085] This step aims to include more detailed information about small lesions in the first image.

[0086] S330, input the updated training sample into the second model to obtain the intermediate lesion segmentation result, compare the intermediate lesion segmentation result with the label in the target sample to optimize the model parameters of the second model until the model parameters meet the set optimization conditions, and obtain the pre-trained second model.

[0087] After the training samples are updated, the updated training samples are input into the second model to obtain the intermediate lesion segmentation results, and the intermediate lesion segmentation results are compared with the labels of the training samples to optimize the model parameters of the second model until the model parameters meet the set optimization conditions, completing the training of the second model and obtaining the pre-trained second model.

[0088] The training samples in the embodiment of the present invention include labels of lesions whose volumes are smaller than a set volume threshold and lesions whose volumes are smaller than the set volume threshold, wherein the labels are that the voxel values ​​of the voxels corresponding to the lesions whose volumes are smaller than the set volume threshold are 1, and the voxel values ​​of other voxels are zero; combined with resampling to improve the resolution of lesions whose volumes are smaller than the set volume threshold, a pre-trained second model with the ability to segment small lesions can be trained.

[0089] Figure 4 Schematic diagram of the structure of the lesion segmentation device provided by an embodiment of the present invention. Figure 4 As shown, the device comprises:

[0090] A first resampling module 110 is configured to resample the first image to obtain a second image, wherein the voxel size in the second image is smaller than the voxel size in the first image, the first image includes at least one first lesion and at least one second lesion, and the volume of the second lesion is smaller than the volume of the first lesion;

[0091] A segmentation module 120, configured to input the first image and the second image into a pre-trained first model and a pre-trained second model, respectively, to obtain a first lesion segmentation result and a second lesion segmentation result;

[0092] A second resampling module 130, configured to resample the second lesion segmentation result to obtain a third lesion segmentation result, wherein the voxel size in the third lesion segmentation result is the same as the voxel size in the first image;

[0093] The result module 140 is used to determine a target lesion segmentation result according to the first lesion segmentation result and the third lesion segmentation result.

[0094] In one embodiment, Figure 5 As shown, the device further includes an image cropping module 100, and the image cropping module 100 is used to:

[0095] Cropping the initial image based on the SUV information to obtain a first image, wherein the SUV information included in the first image is equal to the SUV information included in the initial image;

[0096] The result module 140 is specifically used for:

[0097] Determine two base plate images corresponding to the initial image, wherein the voxel value of each voxel in the base plate images is zero;

[0098] According to the position of the first image in the initial image, adding the first lesion segmentation result to one base plate image to update the first lesion segmentation result, and adding the third lesion segmentation result to another base plate image to update the third lesion segmentation result;

[0099] The union of the updated first lesion segmentation result and the updated third lesion segmentation result is used as the target lesion segmentation result;

[0100] The target lesion segmentation result is displayed.

[0101] In one embodiment, the result module 140 is specifically used to:

[0102] Deleting lesions whose volumes are greater than a set volume threshold from the updated third lesion segmentation result to obtain a fourth lesion segmentation result;

[0103] The target segmentation result is determined according to the updated first lesion segmentation result and the fourth lesion segmentation result.

[0104] In one embodiment, the first resampling module is further configured to:

[0105] The first image is resampled to update the first image, and a voxel size in the updated first image is larger than a voxel size in the second image and smaller than a voxel size in the first image.

[0106] In one embodiment, Figure 6 As shown, the device further includes a normalization module 150, and the normalization module 150 is used to:

[0107] The first image and the second image are normalized respectively based on the SUV information to obtain an updated first image and an updated second image.

[0108] In one embodiment, the initial image is a prostate PET image.

[0109] In one embodiment, the segmentation module is specifically used to:

[0110] Inputting the first image and the second image into a pre-trained first model and a pre-trained second model respectively, so as to obtain a lesion percentage corresponding to each voxel in the first image and a lesion percentage corresponding to each voxel in the second image;

[0111] Determining the first lesion segmentation result according to a voxel distribution range in which the lesion percentage in the first image is greater than a set threshold;

[0112] The second lesion segmentation result is determined according to the voxel distribution range in which the lesion percentage in the second image is greater than a set threshold.

[0113] In one embodiment, the device further comprises a model training module, wherein the model training module is used to:

[0114] Acquire a training sample set, the training sample set comprising at least two training samples, the training samples comprising a label and a lesion whose volume is smaller than a set volume threshold, the label being that a voxel value of a voxel corresponding to a lesion whose volume is smaller than the set volume threshold is 1, and a voxel value of other voxels is zero;

[0115] Resampling each training sample to update each training sample, wherein the voxel size of each updated training sample is the same as the voxel size of the first image;

[0116] The updated training samples are input into the second model to obtain intermediate lesion segmentation results, and the intermediate lesion segmentation results are compared with the labels in the training samples to optimize the model parameters of the second model until the model parameters meet the set optimization conditions to obtain a pre-trained second model.

[0117] The technical solution provided by the embodiment of the present invention obtains a second image with higher spatial resolution by reducing the voxel size of the first image; the first image and the second image are processed respectively by a pre-trained first model and a pre-trained second model to obtain a first lesion segmentation result and a second lesion segmentation result. Since the voxel size corresponding to the second image is smaller than the voxel size corresponding to the first image, the second lesion segmentation result includes some small lesion segmentation results that do not exist in the first lesion segmentation result. Therefore, the second lesion segmentation result is converted from the second accuracy to the first accuracy to obtain a third lesion segmentation result. The target lesion segmentation result corresponding to the union of the first lesion segmentation result and the third lesion segmentation result has high accuracy and no small lesions will be missed.

[0118] The lesion segmentation device provided in the embodiment of the present invention can execute the lesion segmentation method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0119] Figure 7 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0120] like Figure 7As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0121] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0122] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 11 executes the various methods and processes described above, such as a lesion segmentation method.

[0123] In some embodiments, the lesion segmentation method can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the lesion segmentation method described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to perform the lesion segmentation method in any other appropriate manner (e.g., by means of firmware).

[0124] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0125] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0126] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0127] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0128] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0129] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0130] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0131] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A lesion segmentation method, characterized in that: include: Resampling the first image to obtain a second image, wherein a voxel size in the second image is smaller than a voxel size in the first image, the first image includes at least one first lesion and at least one second lesion, and a volume of the second lesion is smaller than a volume of the first lesion; Inputting the first image and the second image into a pre-trained first model and a pre-trained second model respectively to obtain a first lesion segmentation result and a second lesion segmentation result; resampling the second lesion segmentation result to obtain a third lesion segmentation result, wherein the voxel size in the third lesion segmentation result is the same as the voxel size in the first image; A target lesion segmentation result is determined according to the first lesion segmentation result and the third lesion segmentation result.

2. The method according to claim 1, characterized in that Before resampling the first image to obtain the second image, the method further includes: Cropping the initial image based on the SUV information to obtain a first image, wherein the SUV information included in the first image is equal to the SUV information included in the initial image; The determining the target lesion segmentation result according to the first lesion segmentation result and the third lesion segmentation result includes: Determine two base plate images corresponding to the initial image, wherein the voxel value of each voxel in the base plate images is zero; According to the position of the first image in the initial image, adding the first lesion segmentation result to one base plate image to update the first lesion segmentation result, and adding the third lesion segmentation result to another base plate image to update the third lesion segmentation result; The union of the updated first lesion segmentation result and the updated third lesion segmentation result is used as the target lesion segmentation result; The target lesion segmentation result is displayed.

3. The method according to claim 2, characterized in that The step of taking the union of the updated first lesion segmentation result and the updated third lesion segmentation result as the target lesion segmentation result includes: Deleting lesions whose volumes are greater than a set volume threshold from the updated third lesion segmentation result to obtain a fourth lesion segmentation result; The target segmentation result is determined according to the updated first lesion segmentation result and the fourth lesion segmentation result.

4. The method according to claim 1, characterized in that Before inputting the first image and the second image into a pre-trained first model and a pre-trained second model respectively to obtain a first lesion segmentation result and a second lesion segmentation result, the method further includes: The first image is resampled to update the first image, and a voxel size in the updated first image is larger than a voxel size in the second image and smaller than a voxel size in the first image.

5. The method according to claim 1, characterized in that Before inputting the first image and the second image into a pre-trained first model and a pre-trained second model respectively to obtain a first lesion segmentation result and a second lesion segmentation result, the method further includes: The first image and the second image are normalized respectively based on the SUV information to obtain an updated first image and an updated second image.

6. The method according to claim 1, characterized in that The step of inputting the first image and the second image into a pre-trained first model and a pre-trained second model respectively to obtain a first lesion segmentation result and a second lesion segmentation result comprises: Inputting the first image and the second image into a pre-trained first model and a pre-trained second model respectively, so as to obtain a lesion percentage corresponding to each voxel in the first image and a lesion percentage corresponding to each voxel in the second image; Determining the first lesion segmentation result according to a voxel distribution range in which the lesion percentage in the first image is greater than a set threshold; The second lesion segmentation result is determined according to a voxel distribution range in which the lesion percentage in the second image is greater than a set threshold.

7. The method according to claim 1, characterized in that The training of the second model is completed by: Acquire a training sample set, the training sample set comprising at least two training samples, the training samples comprising a label and a lesion whose volume is smaller than a set volume threshold, the label being that a voxel value of a voxel corresponding to a lesion whose volume is smaller than the set volume threshold is 1, and a voxel value of other voxels is zero; Resampling each training sample to update each training sample, wherein the voxel size of each updated training sample is the same as the voxel size of the first image; The updated training samples are input into the second model to obtain intermediate lesion segmentation results, and the intermediate lesion segmentation results are compared with the labels in the training samples to optimize the model parameters of the second model until the model parameters meet the set optimization conditions to obtain a pre-trained second model.

8. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the lesion segmentation method according to any one of claims 1-7.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the lesion segmentation method according to any one of claims 1 to 6 when executed.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the computer program implements the lesion segmentation method according to any one of claims 1-7.

Citation Information

Patent Citations

  • Lung focus identification method and device and realization device

    CN107644225A

  • CT image segmentation method and system based on deep learning

    CN116503607A