CN113763394A - Medical image segmentation control method based on medical risk - Google Patents
Medical image segmentation control method based on medical risk Download PDFInfo
- Publication number
- CN113763394A CN113763394A CN202110973635.XA CN202110973635A CN113763394A CN 113763394 A CN113763394 A CN 113763394A CN 202110973635 A CN202110973635 A CN 202110973635A CN 113763394 A CN113763394 A CN 113763394A
- Authority
- CN
- China
- Prior art keywords
- medical image
- image segmentation
- medical
- control method
- layers
- 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
- 238000003709 image segmentation Methods 0.000 title claims abstract description 70
- 238000000034 method Methods 0.000 title claims abstract description 27
- 230000011218 segmentation Effects 0.000 claims abstract description 38
- 238000012549 training Methods 0.000 claims abstract description 13
- 238000012360 testing method Methods 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 5
- 238000002372 labelling Methods 0.000 claims abstract description 4
- 238000011176 pooling Methods 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 description 13
- 210000001625 seminal vesicle Anatomy 0.000 description 10
- 201000010653 vesiculitis Diseases 0.000 description 10
- 210000000664 rectum Anatomy 0.000 description 9
- 210000002307 prostate Anatomy 0.000 description 7
- 210000000056 organ Anatomy 0.000 description 4
- 206010028980 Neoplasm Diseases 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 210000001519 tissue Anatomy 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000003187 abdominal effect Effects 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000004204 blood vessel Anatomy 0.000 description 1
- 201000011510 cancer Diseases 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000002271 resection Methods 0.000 description 1
Images
Classifications
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/30081—Prostate
-
- 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
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a medical image segmentation control method based on medical risks, which specifically comprises the following steps: s1, acquiring a medical image set to be segmented, and labeling a target area in the medical image set; s2, preprocessing the original data in the medical image set; s3, dividing the medical image set into a training set and a testing set according to a preset proportion; s4, generating a medical image segmentation model, inputting the training set into the medical image segmentation model, and training the medical image segmentation model; and S5, inputting the test set into the trained image segmentation model to obtain a segmented image of the target area. Compared with the prior art, the method has the advantages that the model is always restrained by medical risks, the accuracy of the medical image segmentation result is improved, accurate and safe segmentation is realized, and the like.
Description
Technical Field
The invention relates to the field of medical image segmentation, in particular to a medical image segmentation control method based on medical risks.
Background
In recent years, diseases such as cancer have been on the trend of being less aged and having a high incidence due to factors such as destruction of human living environment, increase in work pressure, and excessively fast pace of life. In clinical treatment of tumor, excision of lesion area is a more common and effective treatment method at present. When a doctor performs a resection operation, the diseased region needs to be completely resected, and meanwhile, the surrounding normal tissues and organs, such as blood vessels, nerves and other organs, need to be prevented from being damaged as much as possible, so that the medical risk is reduced. With the continuous development and progress of computer technology, a great deal of research shows that the medical image is segmented in a full-automatic mode, and great help is provided for doctors in the aspects of diagnosis and treatment. However, since the medical images themselves have high complexity and the shapes of tissues and organs vary from person to person, automatic segmentation of specific targets from the medical images is a difficult task, and the accuracy of segmentation results based on the medical images still needs to be improved at present, and research on the aspect becomes a hot problem in the field of computer vision.
Disclosure of Invention
The present invention is directed to overcome the above-mentioned drawbacks of the prior art, and an object of the present invention is to provide a medical image segmentation control method based on medical risks, which effectively controls medical risks during segmentation, so that a segmentation result with minimized risks can be obtained without reducing segmentation accuracy.
The purpose of the invention can be realized by the following technical scheme:
a medical image segmentation control method based on medical risks specifically comprises the following steps:
s1, acquiring a medical image set to be segmented, and labeling a target area in the medical image set;
s2, preprocessing the original data in the medical image set;
s3, dividing the medical image set into a training set and a testing set according to a preset proportion;
s4, generating a medical image segmentation model, inputting a training set into the medical image segmentation model, and training the medical image segmentation model;
and S5, inputting the test set into the trained image segmentation model to obtain a segmented image of the target area.
The medical image segmentation model is generated based on a U-net network structure.
The step S4 of generating the medical image segmentation model specifically includes the following steps:
s41, constructing coding blocks required by the medical image segmentation model;
s42, decoding blocks required by construction of a medical image segmentation model;
and S43, setting a loss function of the medical image segmentation model, and generating the medical image segmentation model by combining the corresponding coding block and decoding block.
The coding block structure is specifically a four-layer coding block.
Furthermore, each of the four layers of coding blocks includes a convolution layer, a ReLU active layer, and a maximum pooling layer, which are connected in sequence.
Further, the number of convolution layers in each layer of coding block is 2, the number of ReLU active layers is 1, and the number of maximum pooling layers is 1.
The decoding block has a structure specifically including four layers of decoding blocks.
Further, each of the four decoding layers comprises a convolutional layer, an upsampling layer, a ReLU active layer, and a skip connection layer.
Further, the number of convolutional layers in each layer of decoding block is 2, the number of upsampling layers is 1, the number of ReLU active layers is 1, and the number of skip connection layers is 1.
The loss function LsegThe medical risk loss term is combined with a cross entropy function, and the specific formula is as follows:
Lseg＝Lce+Lr
wherein L isceFor a multi-class cross entropy loss function, LrFor the medical risk loss function, y is the standard segmentation map,
is a segmentation graph output by the decoding block, lambda is a hyper-parameter, the corresponding value is automatically learned through the pre-definition or the neural network,
is a process variable.
Compared with the prior art, the invention has the following beneficial effects:
based on the requirements in practical clinical application, the risk minimization thought is introduced on the basis of the classical U-Net model, the loss function is optimized, and the medical risk loss item is added on the basis of the original cross entropy loss function, so that the model is always constrained by the medical risk in the learning process, the accuracy of the medical image segmentation result is effectively improved, and accurate and safe segmentation is realized.
Drawings
FIG. 1 is a schematic structural view of the present invention;
FIG. 2 is a medical image segmentation model specific structure parameter map based on medical risk according to the present invention;
FIG. 3 is a comparison graph of S-risk values of wrongly-divided rectum images in different model segmentation results according to an embodiment of the present invention;
FIG. 4 is a comparison graph of Dice values of wrongly-divided rectal images in different model segmentation results according to an embodiment of the present invention;
FIG. 5 is a comparison graph of S-risk values of wrongly-divided seminal vesicle images in different model segmentation results according to an embodiment of the present invention;
FIG. 6 is a comparison diagram of Dice values of misclassified seminal vesicle images in different model segmentation results according to an embodiment of the present invention.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
Examples
As shown in fig. 1, a medical image segmentation control method based on medical risk specifically includes the following steps:
s1, acquiring a medical image set to be segmented, and labeling a target area in the medical image set;
s2, preprocessing the original data in the medical image set;
s3, dividing the medical image set into a training set and a testing set according to a preset proportion, wherein in the embodiment, the preset proportion is set to be 8: 2;
s4, generating a medical image segmentation model, inputting the training set into the medical image segmentation model, and training the medical image segmentation model;
and S5, inputting the test set into the trained image segmentation model to obtain a segmented image of the target area.
The medical image segmentation model is generated based on a U-net network structure, which is shown in fig. 2.
In this embodiment, the U-net network training adopts Adam algorithm to optimize the loss function, the size of batch size is set to 8, epoch is set to 300, and the learning rate is set to 0.00001.
In this embodiment, in the data preprocessing process of step S2, in order to ensure that the images have the same size and resolution, the spatial resolutions of the original slice image and the annotated image are uniformly resampled to 256 × 256.
The step S4 of generating the medical image segmentation model specifically includes the following steps:
s41, constructing coding blocks required by the medical image segmentation model;
s42, decoding blocks required by construction of a medical image segmentation model;
and S43, setting a loss function of the medical image segmentation model, and generating the medical image segmentation model by combining the corresponding coding block and decoding block.
The coding block structure is specifically a four-layer coding block, and each layer of coding block in the four-layer coding block comprises a convolution layer, a ReLU active layer and a maximum pooling layer which are sequentially connected.
The number of the convolution layers in each layer of coding block is 2, the number of the ReLU active layers is 1, and the number of the maximum pooling layers is 1.
The decoding block structure is specifically four layers of decoding blocks, wherein each layer of decoding block in the four layers of decoding blocks comprises a convolution layer, an up-sampling layer, a ReLU activation layer and a jump connection layer.
The number of convolution layers in each layer of decoding block is 2, the number of up-sampling layers is 1, the number of ReLU active layers is 1, and the number of jump connection layers is 1.
Loss function LsegThe medical risk loss term is combined with a cross entropy function, and the specific formula is as follows:
Lseg＝Lce+Lr
wherein L isceFor a multi-class cross entropy loss function, LrFor the medical risk loss function, y is the standard segmentation map,
is a segmentation graph output by the decoding block, lambda is a hyper-parameter, the corresponding value is automatically learned through the pre-definition or the neural network,
is a process variable.
In specific implementation, taking prostate segmentation as an example, the data set is obtained by imaging an abdominal scan with a siemens 3.0T magnetic resonance scanner using different combinations of radio-frequency and pulses, and is a transverse T2 weighted MR image. The data set contains 82 sample data, each sample data contains about 24 image slices in DICOM format and corresponding labeled images manually segmented by a professional doctor as a segmentation golden standard, and the standard segmentation images are in NII format. The organs contained in the images are the prostate, seminal vesicle and rectum. The segmentation result is evaluated by a segmentation accuracy (Dice) and a medical risk value (S-risk).
Through multiple experiments, in the segmentation result given by the original U-net algorithm, the condition that the rectum is cut by mistake when the prostate of the target region is segmented by 13 images in a test set on average is tested. By using the medical image segmentation model of the invention, the test set has 9 images on average, and the rectum is cut by mistake when the prostate of the target area is segmented. Aiming at 13 images of the rectum with the miscut in the classic U-Net algorithm, the medical image segmentation model completely avoids the miscut of 6 images, and greatly reduces the miscut range of the other 6 images. In addition, 2 new images with rectal cross-cuts appear in the medical image segmentation model of the invention, but the cross-cuts range is very small and are respectively 1 pixel and 2 pixels.
As shown in FIG. 3, under the S-risk evaluation criterion, the medical image segmentation model of the invention obtains lower risk values than the classical U-Net algorithm in 14 out of 15 wrong rectum images in total, namely, the medical image segmentation model of the invention can better avoid or reduce the risk of wrong rectum division in segmentation. As shown in FIG. 4, the Dice coefficient value of the segmentation result obtained by the medical image segmentation model of the invention is basically coincident with the Dice coefficient value of the segmentation result obtained by the classic U-Net algorithm or higher than the Dice coefficient value of the segmentation result obtained by the classic U-Net algorithm, which shows that the medical image segmentation model of the invention can achieve at least the same effect as the U-Net algorithm in the segmentation accuracy.
Aiming at the situation of seminal vesicle miscut, in the segmentation result of the test set, the U-Net algorithm performs miscut on 36 images to obtain the prostate of the target region, and the medical image segmentation model performs miscut on 35 images to obtain the prostate of the target region. In 36 images with miscut seminal vesicle in the classic U-Net algorithm, the medical image segmentation model completely avoids the miscut of 1 image, and greatly reduces the miscut range of the other 24 images. In addition, 1 new picture with seminal vesicle miscut appears in the medical image segmentation model of the invention, but the miscut range is very small and is only 1 pixel.
As shown in fig. 5, the Dice coefficient value of the segmentation result obtained by the medical image segmentation model of the present invention substantially coincides with the Dice coefficient value of the segmentation result obtained by the classical U-Net algorithm, and there is a large difference between the two values only in the individual images (e.g. 9 th, 23 th, and 31 th images), but there is no single bias, so the overall segmentation accuracy of the medical image segmentation model of the present invention is substantially equal to that of the U-Net algorithm. As shown in FIG. 6, under the S-risk evaluation criterion, the medical image segmentation model of the invention obtains a lower risk value than the classic U-Net algorithm in 26 out of 37 misclassified seminal vesicle images in total, i.e. the medical image segmentation model of the invention can better avoid or reduce the risk of misclassified seminal vesicle in segmentation.
As shown in table 1, the average scores of all miscut rectum (15 images) and all miscut seminal vesicle (37 images) over Dice coefficient and R-risk were calculated, respectively, table 1 is specified below:
TABLE 1 comparison of average segmentation effect of misclassified images in segmentation results of different models
It can be seen that the risk values for the rectum and seminal vesicle decrease from 0.0049 to 0.0009 and from 0.384 to 0.299 by 82% and 22%, respectively, while the accuracy for the prostate segmentation increases by about two percentage points and decreases by about 0.1 percentage point, respectively, and thus it can be seen that the medical image segmentation model of the present invention can effectively reduce the medical risk during the segmentation process while hardly decreasing the segmentation accuracy.
In addition, it should be noted that the specific embodiments described in the present specification may have different names, and the above descriptions in the present specification are only illustrations of the structures of the present invention. All equivalent or simple changes in the structure, characteristics and principles of the invention are included in the protection scope of the invention. Various modifications or additions may be made to the described embodiments or methods may be similarly employed by those skilled in the art without departing from the scope of the invention as defined in the appending claims.
Claims (10)
1. A medical image segmentation control method based on medical risks is characterized by comprising the following steps:
s1, acquiring a medical image set to be segmented, and labeling a target area in the medical image set;
s2, preprocessing the original data in the medical image set;
s3, dividing the medical image set into a training set and a testing set according to a preset proportion;
s4, generating a medical image segmentation model, inputting a training set into the medical image segmentation model, and training the medical image segmentation model;
and S5, inputting the test set into the trained image segmentation model to obtain a segmented image of the target area.
2. The medical image segmentation control method based on medical risk according to claim 1, wherein the medical image segmentation model is generated based on a U-net network structure.
3. The medical image segmentation control method based on medical risk according to claim 1, wherein the step S4 of generating the medical image segmentation model specifically includes the steps of:
s41, constructing coding blocks required by the medical image segmentation model;
s42, decoding blocks required by construction of a medical image segmentation model;
and S43, setting a loss function of the medical image segmentation model, and generating the medical image segmentation model by combining the corresponding coding block and decoding block.
4. The medical image segmentation control method based on medical risk according to claim 3, wherein the structure of the coding blocks is four layers of coding blocks.
5. The medical image segmentation control method based on medical risk according to claim 4, wherein each of the four layers of coding blocks comprises a convolution layer, a ReLU activation layer and a max pooling layer which are connected in sequence.
6. The medical image segmentation control method based on medical risk according to claim 5, wherein the number of convolution layers in each layer coding block is 2, the number of ReLU active layers is 1, and the number of maximum pooling layers is 1.
7. The medical risk-based medical image segmentation control method according to claim 3, wherein the decoding block has a structure of four layers.
8. The medical risk-based medical image segmentation control method according to claim 7, wherein each of the four decoding layers comprises a convolutional layer, an upsampling layer, a ReLU activation layer, and a skip connection layer.
9. The medical risk-based medical image segmentation control method according to claim 8, wherein the number of convolutional layers in each layer of decoding block is 2, the number of upsampling layers is 1, the number of ReLU active layers is 1, and the number of skip connection layers is 1.
10. The medical risk-based medical image segmentation control method according to claim 3, wherein the loss function LsegThe medical risk loss term is combined with a cross entropy function, and the specific formula is as follows:
Lseg＝Lce+Lr
wherein L isceFor a multi-class cross entropy loss function, LrFor the medical risk loss function, y is the standard segmentation map,
is a segmentation graph output by the decoding block, lambda is a hyper-parameter,
is a process variable.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110973635.XA CN113763394B (en) | 2021-08-24 | 2021-08-24 | Medical image segmentation control method based on medical risks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110973635.XA CN113763394B (en) | 2021-08-24 | 2021-08-24 | Medical image segmentation control method based on medical risks |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113763394A true CN113763394A (en) | 2021-12-07 |
| CN113763394B CN113763394B (en) | 2024-03-29 |
Family
ID=78790973
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110973635.XA Active CN113763394B (en) | 2021-08-24 | 2021-08-24 | Medical image segmentation control method based on medical risks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113763394B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190030371A1 (en) * | 2017-07-28 | 2019-01-31 | Elekta, Inc. | Automated image segmentation using dcnn such as for radiation therapy |
| US20190223725A1 (en) * | 2018-01-25 | 2019-07-25 | Siemens Healthcare Gmbh | Machine Learning-based Segmentation for Cardiac Medical Imaging |
| US20200074637A1 (en) * | 2018-08-28 | 2020-03-05 | International Business Machines Corporation | 3d segmentation with exponential logarithmic loss for highly unbalanced object sizes |
| WO2021017372A1 (en) * | 2019-08-01 | 2021-02-04 | 中国科学院深圳先进技术研究院 | Medical image segmentation method and system based on generative adversarial network, and electronic equipment |
| CN113129309A (en) * | 2021-03-04 | 2021-07-16 | 同济大学 | Medical image semi-supervised segmentation system based on object context consistency constraint |
-
2021
- 2021-08-24 CN CN202110973635.XA patent/CN113763394B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190030371A1 (en) * | 2017-07-28 | 2019-01-31 | Elekta, Inc. | Automated image segmentation using dcnn such as for radiation therapy |
| US20190223725A1 (en) * | 2018-01-25 | 2019-07-25 | Siemens Healthcare Gmbh | Machine Learning-based Segmentation for Cardiac Medical Imaging |
| US20200074637A1 (en) * | 2018-08-28 | 2020-03-05 | International Business Machines Corporation | 3d segmentation with exponential logarithmic loss for highly unbalanced object sizes |
| WO2021017372A1 (en) * | 2019-08-01 | 2021-02-04 | 中国科学院深圳先进技术研究院 | Medical image segmentation method and system based on generative adversarial network, and electronic equipment |
| CN113129309A (en) * | 2021-03-04 | 2021-07-16 | 同济大学 | Medical image semi-supervised segmentation system based on object context consistency constraint |
Non-Patent Citations (1)
| Title |
|---|
| 石军;赵敏帆;薛旭东;郝晓宇;金旭;安虹;张红雁;: "2D级联CNN模型的放疗危及器官自动分割", 中国图象图形学报, no. 10 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113763394B (en) | 2024-03-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109035255B (en) | Method for segmenting aorta with interlayer in CT image based on convolutional neural network | |
| CN109598722B (en) | Image analysis method based on recurrent neural network | |
| US7792339B2 (en) | Method and apparatus for intracerebral hemorrhage lesion segmentation | |
| CN110070540B (en) | Image generation method and device, computer equipment and storage medium | |
| Nikan et al. | PWD-3DNet: a deep learning-based fully-automated segmentation of multiple structures on temporal bone CT scans | |
| CN111640100A (en) | Tumor image processing method and device, electronic equipment and storage medium | |
| CN110880366B (en) | Medical image processing system | |
| US20110123074A1 (en) | Systems and methods for suppressing artificial objects in medical images | |
| CN110264449B (en) | Quality control method of medical radiological image chest radiography based on artificial intelligence | |
| CN109934829B (en) | Liver segmentation method based on three-dimensional graph segmentation algorithm | |
| CN112036506A (en) | Image recognition method and related device and equipment | |
| CN113239755B (en) | Medical hyperspectral image classification method based on space-spectrum fusion deep learning | |
| CN116503607B (en) | CT image segmentation method and system based on deep learning | |
| CN114299072A (en) | Artificial intelligence-based anatomy variation identification prompting method and system | |
| CN112686336A (en) | Burn surface of a wound degree of depth classification system based on neural network | |
| US20130069945A1 (en) | Method for displaying the information contained in three-dimensional images of the heart | |
| CN113223028A (en) | Multi-modal liver tumor segmentation method based on MR and CT | |
| CN106557767B (en) | A kind of method of ROI region in determining interventional imaging | |
| CN104915989B (en) | Blood vessel three-dimensional dividing method based on CT images | |
| Xu et al. | AI-CHD: an AI-based framework for cost-effective surgical telementoring of congenital heart disease | |
| CN113763394B (en) | Medical image segmentation control method based on medical risks | |
| CN116229067A (en) | Channel attention-based liver cell cancer CT image segmentation method | |
| CN116128819A (en) | Image processing method, model training method and device and electronic equipment | |
| CN109685803A (en) | A kind of left ventricle image partition method, device, equipment and storage medium | |
| CN113379682B (en) | Heart MRI image coupling level set segmentation method and system |
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 |