CN115311453A - Automatic hippocampus segmentation method and system based on V-Net and combined loss function - Google Patents
Automatic hippocampus segmentation method and system based on V-Net and combined loss function Download PDFInfo
- Publication number
- CN115311453A CN115311453A CN202210739917.8A CN202210739917A CN115311453A CN 115311453 A CN115311453 A CN 115311453A CN 202210739917 A CN202210739917 A CN 202210739917A CN 115311453 A CN115311453 A CN 115311453A
- Authority
- CN
- China
- Prior art keywords
- hippocampus
- net
- image data
- loss function
- magnetic resonance
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10088—Magnetic resonance imaging [MRI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20092—Interactive image processing based on input by user
- G06T2207/20104—Interactive definition of region of interest [ROI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30016—Brain
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/80—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in fisheries management
- Y02A40/81—Aquaculture, e.g. of fish
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention belongs to the technical field of brain image analysis, and particularly relates to a hippocampus automatic segmentation method and system based on V-Net and a combined loss function, wherein the data is preprocessed by acquiring nuclear magnetic resonance image data of a subject; aiming at the preprocessed nuclear magnetic resonance image data of the testee, a hippocampus mask is obtained by utilizing a hippocampus automatic segmentation model so as to realize the automatic segmentation of the hippocampus of the testee, wherein the hippocampus automatic segmentation model adopts a trained V-Net network model structure, the V-Net network model supplements loss information by mapping the superposition characteristics of a compression path to a decompression path, and training is carried out by utilizing a combined loss function containing the maximization of a Dice coefficient. According to the invention, the image characteristics of the hippocampus are mined through the V-Net and the combined loss function, the automatic division efficiency and accuracy of the hippocampus are improved, and the application in practical scenes is facilitated.
Description
Technical Field
The invention belongs to the technical field of brain image analysis, and particularly relates to a hippocampus automatic segmentation method and system based on V-Net and a combined loss function.
Background
Hippocampus (hippopopus) is located between the thalamus and medial temporal lobe of the brain, is responsible for memory storage and retrieval, and is one of the most severely affected brain regions by Alzheimer's Disease (AD). Since hippocampal atrophy is among the early overt symptoms of AD patients, the volume and morphological features of the hippocampal region can be used as important biomarkers to aid in the diagnosis of AD. In recent years, with the development of the field of artificial intelligence, various algorithm models have been gradually used in research and analysis of neuroimaging data. The hippocampus is automatically and accurately segmented by applying an artificial intelligence method, which is helpful for doctors to quickly diagnose the illness state of AD and other related diseases clinically.
Due to the irregularity of the anatomical structure of the hippocampus, the problems of low contrast, fuzzy edge and the like exist, so that the difficulty of partitioning the hippocampus is high. How to apply an advanced artificial intelligence algorithm and improve the efficiency and the accuracy of the automatic division of the hippocampus is one of the research directions concerned by researchers.
Disclosure of Invention
Therefore, the invention provides a method and a system for automatically segmenting the hippocampus based on the V-Net and the combined loss function, which are used for mining the image characteristics of the hippocampus through the V-Net and the combined loss function, improving the automatic segmentation efficiency and accuracy of the hippocampus and facilitating the application of actual scenes.
According to the design scheme provided by the invention, the invention provides a hippocampus automatic segmentation method based on V-Net and a combined loss function, which comprises the following contents:
acquiring nuclear magnetic resonance image data of a subject, and preprocessing the data;
aiming at the preprocessed nuclear magnetic resonance image data of the testee, a hippocampus mask is obtained by utilizing a hippocampus automatic segmentation model so as to realize the automatic segmentation of the hippocampus of the testee, wherein the hippocampus automatic segmentation model adopts a trained V-Net network model structure, the V-Net network model supplements loss information by mapping the superposition characteristics of compression paths to decompression paths, and the V-Net network model is trained by utilizing a combinability loss function containing the maximization of a Dice coefficient.
As the automatic segmentation method of the hippocampus based on V-Net and combined loss function in the invention, further, the original medical image data of the subject is collected by using a magnetic resonance imaging method, and nuclear magnetic resonance image data is obtained by using a weighted imaging algorithm, wherein the magnetic resonance image data at least comprises the following components: 1.5T of T1-weighted imaging sequences.
As the automatic division method of the hippocampus based on the V-Net and the combined loss function, the preprocessing operation at least comprises the following steps: the method comprises the steps of NIFTI format unified conversion processing of magnetic resonance image data, registration processing, skull removal, bias field correction, brightness range limitation by using a threshold value, background removal, data enhancement and cutting processing.
As the method for automatically segmenting the hippocampus based on the V-Net and the combined loss function, the method further comprises the following steps in preprocessing the magnetic resonance image data: and carrying out region-of-interest labeling processing on the image data, and acquiring the position of the hippocampus in the image data by using the labeling processing.
As the automatic hippocampus segmentation method based on the V-Net and combined loss function, the invention further utilizes an Alzheimer disease ADNI data set as a training sample to train the automatic hippocampus segmentation model, wherein the training sample comprises: a T1 weighted imaging sequence of 1.5T and a region of interest labeled in each subject's image data.
As the automatic segmentation method of the hippocampus based on the V-Net and the combined loss function, further, the V-Net network model structure comprises a compression path consisting of an encoder and a decompression path consisting of a decoder, wherein different convolutional layers are utilized in the compression path, and a uniform stride size is applied to the convolutional cores on the same convolutional layer, and input data sequentially carry out convolution operation along the convolutional layers in the compression path; in the decompression path, the compressed image features are restored to the original size by the transposed convolutional layer corresponding to the compression path, and the size is controlled by the padding operation in the transposed convolutional layer.
As the automatic partition method of the hippocampus based on the V-Net and the combined loss function, further, the V-Net network model adopts a ResNet short circuit connection mode, and in the convolution stage and the transposition convolution stage of a convolution compression path and a decompression path, the superposition characteristics of the compression path are mapped to the decompression path through a learning residual error function so as to supplement the loss information in the learning process.
The invention relates to a hippocampus automatic segmentation method based on V-Net and a combined loss function, and further, the combined loss function is expressed as:
therein, loss Dice Loss function, loss, representing maximization of Dice coefficient Cross A classification loss function in the segmentation is represented,
M i representing the number of pixels belonging to the region of interest i in the current training sample label, M representing the total number of pixels in the current training sample label, y i Indicating that the true label value in the training sample is the one-hot encoding of the region of interest i,
representing the network segmentation prediction result as the probability value of the interest i, W i To correspond toAnd (4) weighting.
Further, the present invention provides an automatic hippocampus segmentation system based on V-Net and combined loss function, comprising: a data acquisition module and a segmentation processing module, wherein,
the data acquisition module is used for acquiring nuclear magnetic resonance image data of a subject and carrying out preprocessing operation on the data;
and the segmentation processing module is used for acquiring a hippocampus mask by utilizing a hippocampus automatic segmentation model aiming at the preprocessed nuclear magnetic resonance image data of the testee so as to realize the automatic segmentation of the hippocampus of the testee, wherein the hippocampus automatic segmentation model adopts a trained V-Net network model structure, the V-Net network model supplements loss information by mapping the superposition characteristics of a compression path to a decompression path, and the V-Net network model is trained by utilizing a combined loss function containing the maximization of the Dice coefficient.
The invention has the beneficial effects that:
according to the invention, through acquiring the magnetic resonance image data of a subject, a series of data preprocessing operations such as format unified conversion, registration, skull removal, bias field correction, brightness range limitation, background removal, data enhancement, cutting and the like are carried out, the magnetic resonance image characteristics are mined by utilizing an encoder-decoder structure in a hippocampus automatic segmentation model based on V-Net and a combined loss function, the model training efficiency is improved by utilizing the combined loss function of dynamic weight, the automatic segmentation of the hippocampus is realized, the hippocampus segmentation efficiency and accuracy are improved, and the application in practical scenes is facilitated.
Description of the drawings:
FIG. 1 is a schematic diagram of an automatic segmentation process of a hippocampus based on V-Net and a combined loss function in an embodiment;
FIG. 2 is a schematic diagram of an automatic segmentation model of the hippocampus in the embodiment.
The specific implementation mode is as follows:
in order to make the objects, technical solutions and advantages of the present invention clearer and more obvious, the present invention is further described in detail below with reference to the accompanying drawings and technical solutions.
The embodiment of the invention provides a hippocampus automatic segmentation method based on V-Net and a combined loss function, which comprises the following steps: acquiring nuclear magnetic resonance image data of a subject, and preprocessing the data; aiming at the preprocessed nuclear magnetic resonance image data of the testee, a hippocampus mask is obtained by utilizing a hippocampus automatic segmentation model so as to realize the automatic segmentation of the hippocampus of the testee, wherein the hippocampus automatic segmentation model adopts a trained V-Net network model structure, the V-Net network model supplements loss information by mapping the superposition characteristics of compression paths to decompression paths, and the V-Net network model is trained by utilizing a combinability loss function containing the maximization of a Dice coefficient.
Aiming at nuclear magnetic resonance image data of a subject, data preprocessing operations are carried out by adopting registration, limiting a brightness range through a threshold value, removing a background and the like, and hidden information of the 3D image data is mined through an end-to-end means by utilizing a hippocampus automatic segmentation model based on V-Net and a combined loss function, so that a hippocampus region in the image data is automatically identified through an artificial intelligence algorithm, and a doctor can conveniently and rapidly diagnose the illness state of related diseases such as AD and the like clinically.
Further, in this embodiment, a magnetic resonance imaging method is used to acquire raw medical image data of a subject, and a weighted imaging algorithm is used to acquire magnetic resonance image data, where the magnetic resonance image data at least includes: 1.5T of T1-weighted imaging sequences. The original medical image data of the testee acquired by the conventional magnetic resonance imaging method needs to be subjected to a series of preprocessing operations, including preprocessing operations of uniformly converting a data format into NIFTI by using software such as python, performing registration by using software such as FSL, removing the skull by using modules such as SPM or BET, correcting a bias field, limiting a brightness range through a threshold value, removing a background, enhancing data, cutting into uniform size and the like. In preprocessing the magnetic resonance image data, the method further comprises: and carrying out region-of-interest labeling processing on the image data, and acquiring the position of the hippocampus in the image data by using the labeling processing. The method can utilize an Alzheimer's disease ADNI data set as a training sample to train the hippocampus automatic segmentation model, wherein the training sample comprises: a T1 weighted imaging sequence of 1.5T and a region of interest labeled in each subject's image data.
In the embodiment of the scheme, further, the V-Net network model structure comprises a compression path composed of an encoder and a decompression path composed of a decoder, wherein different convolutional layers are utilized in the compression path, a uniform stride size is applied to convolutional cores on the same convolutional layer, and input data sequentially carries out convolution operation along the convolutional layers in the compression path; in the decompression path, the compressed image features are restored to the original size by the transposed convolutional layer corresponding to the compression path, and the size is controlled by the padding operation in the transposed convolutional layer. Furthermore, the V-Net network model adopts a ResNet short circuit connection mode, and in the convolution stage and the transposition convolution stage of the convolution compression path and the decompression path, the compression path superposition characteristics are mapped to the decompression path through a learning residual error function so as to supplement loss information in the learning process.
The automatic segmentation model of the hippocampus based on V-Net and combined loss functions can be built by a framework such as tensierflow or pytorch of python software, see fig. 2, aiming at extracting features from the image data by performing convolution operations based on a conventional 3D convolutional neural network, and reducing its resolution by using an appropriate step at the end of each phase. The entire V-Net model includes two processes, with the compression path on the left and the decompression path on the right. The compression path process can be viewed as distinct stages, running at different resolutions, where each stage includes one to three convolutional layers, the input to each stage is used in the convolutional layer and through a non-linear process, the output is obtained by adding to the last convolutional layer of the stage so that the residual function can be learned. The convolution performed in each stage uses a volume kernel having a size of n × n × n voxels, the resolution of which decreases progressively as the data progresses through the compression path through the different stages, which are performed by convolution using an m × m × m voxel wide kernel applying a step m. During the decompression path, the image features can be restored to the original size, and the padding is adopted during the convolution process to realize size control.
The automatic hippocampus segmentation model based on the V-Net and the combined loss function borrows the idea that the traditional U-Net network superposes feature mapping from a compression path so as to supplement loss information, but the biggest difference with the U-Net is that in each stage, the V-Net adopts a ResNet short circuit connection mode, namely, a residual error module is introduced into an original network model, namely, a residual error function is learned in a convolution stage.
In addition, the hippocampus automatic segmentation analysis model in the present application utilizes a combined loss function based on dynamic weight, and a specific formula can be designed as follows:
wherein M is i Representing the number of pixels belonging to the region of interest i in the current training sample label, M representing the total number of pixels in the current training sample label, y i Indicating that the true label value in the training sample is the one-hot encoding of the region of interest i,
representing the network segmentation prediction result as the probability value of the interest i, W i Are the corresponding weights. By the combined loss function, the segmentation effect of the hippocampus is effectively improved in the model training stage.
Further, based on the foregoing method, an embodiment of the present invention further provides an automatic hippocampus segmentation system based on V-Net and combined loss function, including: a data acquisition module and a segmentation processing module, wherein,
the data acquisition module is used for acquiring nuclear magnetic resonance image data of a subject and carrying out preprocessing operation on the data;
and the segmentation processing module is used for acquiring a hippocampus mask by utilizing a hippocampus automatic segmentation model aiming at the preprocessed nuclear magnetic resonance image data of the testee so as to realize the automatic segmentation of the hippocampus of the testee, wherein the hippocampus automatic segmentation model adopts a trained V-Net network model structure, the V-Net network model supplements loss information by mapping the superposition characteristics of a compression path to a decompression path, and the V-Net network model is trained by utilizing a combined loss function containing the maximization of the Dice coefficient.
Unless specifically stated otherwise, the relative steps, numerical expressions, and values of the components and steps set forth in these embodiments do not limit the scope of the present invention.
In the present specification, the embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The elements of the various examples and method steps described in connection with the embodiments disclosed herein may be embodied in electronic hardware, computer software, or combinations of both, and the components and steps of the examples have been described in a functional generic sense in the foregoing description for clarity of hardware and software interchangeability. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
Those skilled in the art will appreciate that all or part of the steps of the above methods may be implemented by instructing the relevant hardware through a program, which may be stored in a computer-readable storage medium, such as: a read-only memory, a magnetic or optical disk, or the like. Alternatively, all or part of the steps of the foregoing embodiments may also be implemented by using one or more integrated circuits, and accordingly, each module/unit in the foregoing embodiments may be implemented in the form of hardware, and may also be implemented in the form of a software functional module. The present invention is not limited to any specific form of combination of hardware and software.
Finally, it should be noted that: the above-mentioned embodiments are only specific embodiments of the present invention, which are used for illustrating the technical solutions of the present invention and not for limiting the same, and the protection scope of the present invention is not limited thereto, although the present invention is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: any person skilled in the art can modify or easily conceive the technical solutions described in the foregoing embodiments or equivalent substitutes for some technical features within the technical scope of the present disclosure; such modifications, changes or substitutions do not depart from the spirit and scope of the embodiments of the present invention, and they should be construed as being included therein. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. A method for automatically segmenting a hippocampus based on a V-Net and a combined loss function is characterized by comprising the following steps:
acquiring nuclear magnetic resonance image data of a subject, and preprocessing the data;
aiming at the preprocessed nuclear magnetic resonance image data of the testee, a hippocampus mask is obtained by utilizing a hippocampus automatic segmentation model so as to realize the automatic segmentation of the hippocampus of the testee, wherein the hippocampus automatic segmentation model adopts a trained V-Net network model structure, the V-Net network model supplements loss information by mapping the superposition characteristics of compression paths to decompression paths, and the V-Net network model is trained by utilizing a combinability loss function containing the maximization of a Dice coefficient.
2. The method of claim 1, wherein the original medical image data of the subject is acquired by a magnetic resonance imaging method, and the magnetic resonance image data is acquired by a weighted imaging algorithm, the magnetic resonance image data at least comprises: 1.5T of T1-weighted imaging sequences.
3. The method for automatically segmenting hippocampus based on V-Net and combined loss functions according to claim 1 or 2, wherein the preprocessing operation comprises at least: the method comprises the steps of NIFTI format unified conversion processing of magnetic resonance image data, registration processing, skull removal, bias field correction, brightness range limitation by using a threshold value, background removal, data enhancement and cutting processing.
4. The method for automatically segmenting hippocampus based on V-Net and combined loss function according to claim 3, further comprising, in preprocessing magnetic resonance image data: and carrying out region-of-interest labeling processing on the image data, and acquiring the position of the hippocampus in the image data by using the labeling processing.
5. The method for automatically segmenting the hippocampus based on the V-Net and combined loss function according to claim 1, wherein the Alzheimer's disease ADNI data set is used as a training sample to train the hippocampus automatic segmentation model, wherein the training sample comprises: a T1 weighted imaging sequence of 1.5T and a region of interest labeled in the image data of each subject.
6. The method for automatically segmenting hippocampus based on V-Net and combined loss functions according to claim 1 or 5, wherein the V-Net network model structure comprises a compression path composed of an encoder and a decompression path composed of a decoder, wherein the convolution operation is performed by utilizing different convolution layers in the compression path and applying a uniform step size to the convolution kernel on the same convolution layer, and the input data sequentially follows the convolution layers in the compression path; in the decompression path, the compressed image features are restored to the original size by the transposed convolutional layer corresponding to the compression path, and the size is controlled by the padding operation in the transposed convolutional layer.
7. The hippocampus automatic segmentation method based on V-Net and combined loss function according to claim 1, wherein the V-Net network model adopts ResNet short circuit connection mode, and maps the compression path superposition characteristics to the decompression path through learning residual function in convolution stage and transposition convolution stage of convolution compression path and decompression path to supplement loss information in the learning process.
8. The hippocampus automatic segmentation method based on V-Net and combined loss functions according to claim 1, wherein the combined loss function is expressed as:
therein, loss Dice Loss function, loss, representing maximization of Dice coefficient Cross A classification loss function in the segmentation is represented,
M i representing the number of pixels belonging to the region of interest i in the current training sample label, M representing the total number of pixels in the current training sample label, y i Indicating that the true label value in the training sample is the one-hot encoding of the region of interest i,
representing the network segmentation prediction result as the probability value of the region of interest i, W i Are corresponding weights.
9. An automatic hippocampus segmentation system based on V-Net and combined loss functions, comprising: a data acquisition module and a segmentation processing module, wherein,
the data acquisition module is used for acquiring nuclear magnetic resonance image data of a subject and carrying out preprocessing operation on the data;
and the segmentation processing module is used for acquiring a hippocampus mask by utilizing a hippocampus automatic segmentation model aiming at the preprocessed nuclear magnetic resonance image data of the testee so as to realize the automatic segmentation of the hippocampus of the testee, wherein the hippocampus automatic segmentation model adopts a trained V-Net network model structure, the V-Net network model supplements loss information by mapping the superposition characteristics of a compression path to a decompression path, and the V-Net network model is trained by utilizing a combined loss function containing the maximization of the Dice coefficient.
10. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the method steps of any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210739917.8A CN115311453B (en) | 2022-06-27 | 2022-06-27 | Automatic sea horse segmentation method and system based on V-Net and combined loss function |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210739917.8A CN115311453B (en) | 2022-06-27 | 2022-06-27 | Automatic sea horse segmentation method and system based on V-Net and combined loss function |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115311453A true CN115311453A (en) | 2022-11-08 |
| CN115311453B CN115311453B (en) | 2023-06-27 |
Family
ID=83855014
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210739917.8A Active CN115311453B (en) | 2022-06-27 | 2022-06-27 | Automatic sea horse segmentation method and system based on V-Net and combined loss function |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115311453B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3611699A1 (en) * | 2018-08-14 | 2020-02-19 | Siemens Healthcare GmbH | Image segmentation using deep learning techniques |
| CN110969626A (en) * | 2019-11-27 | 2020-04-07 | 西南交通大学 | Method for extracting hippocampus of human brain nuclear magnetic resonance image based on 3D neural network |
| CN111354002A (en) * | 2020-02-07 | 2020-06-30 | 天津大学 | Kidney and kidney tumor segmentation method based on deep neural network |
-
2022
- 2022-06-27 CN CN202210739917.8A patent/CN115311453B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3611699A1 (en) * | 2018-08-14 | 2020-02-19 | Siemens Healthcare GmbH | Image segmentation using deep learning techniques |
| CN110969626A (en) * | 2019-11-27 | 2020-04-07 | 西南交通大学 | Method for extracting hippocampus of human brain nuclear magnetic resonance image based on 3D neural network |
| CN111354002A (en) * | 2020-02-07 | 2020-06-30 | 天津大学 | Kidney and kidney tumor segmentation method based on deep neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115311453B (en) | 2023-06-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114581662B (en) | Brain tumor image segmentation method, system, device and storage medium | |
| CN110265141B (en) | Computer-aided diagnosis method for liver tumor CT image | |
| CN110930416A (en) | MRI image prostate segmentation method based on U-shaped network | |
| CN112785593B (en) | Brain image segmentation method based on deep learning | |
| CN112862805B (en) | Automatic auditory neuroma image segmentation method and system | |
| CN112819914A (en) | PET image processing method | |
| CN112669248A (en) | Hyperspectral and panchromatic image fusion method based on CNN and Laplacian pyramid | |
| CN115375711A (en) | Image segmentation method of global context attention network based on multi-scale fusion | |
| CN114723698A (en) | Cerebrovascular image segmentation method based on multi-scale attention network | |
| CN114998154A (en) | Low-dose CT image denoising method based on transformer and multi-scale features | |
| CN113436128B (en) | Dual-discriminator multi-mode MR image fusion method, system and terminal | |
| CN112750131B (en) | Pelvis nuclear magnetic resonance image musculoskeletal segmentation method based on scale and sequence relation | |
| CN113269774B (en) | Parkinson disease classification and lesion region labeling method of MRI (magnetic resonance imaging) image | |
| Liu | Retinal vessel segmentation based on fully convolutional networks | |
| Shi et al. | Dual dense context-aware network for hippocampal segmentation | |
| CN114066908A (en) | Method and system for brain tumor image segmentation | |
| CN111784652B (en) | MRI (magnetic resonance imaging) segmentation method based on reinforcement learning multi-scale neural network | |
| CN115311453B (en) | Automatic sea horse segmentation method and system based on V-Net and combined loss function | |
| CN115496732A (en) | Semi-supervised heart semantic segmentation algorithm | |
| CN115294023A (en) | Liver tumor automatic segmentation method and device | |
| CN114331996A (en) | Medical image classification method and system based on self-coding decoder | |
| CN114581459A (en) | Improved 3D U-Net model-based segmentation method for image region of interest of preschool child lung | |
| CN111932486A (en) | Brain glioma segmentation method based on 3D convolutional neural network | |
| CN117352120B (en) | GPT-based intelligent self-generation method, device and equipment for knee joint lesion diagnosis | |
| CN117994520A (en) | White matter lesion image segmentation method, system and storage medium based on WMH-Net model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |