CN113450363B - Meta-learning cell nucleus segmentation system and method based on label correction - Google Patents
Meta-learning cell nucleus segmentation system and method based on label correction Download PDFInfo
- Publication number
- CN113450363B CN113450363B CN202110651067.1A CN202110651067A CN113450363B CN 113450363 B CN113450363 B CN 113450363B CN 202110651067 A CN202110651067 A CN 202110651067A CN 113450363 B CN113450363 B CN 113450363B
- Authority
- CN
- China
- Prior art keywords
- correction
- network
- label
- segmentation
- noise
- 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.)
- Active
Links
- 238000012937 correction Methods 0.000 title claims abstract description 87
- 230000011218 segmentation Effects 0.000 title claims abstract description 86
- 238000000034 method Methods 0.000 title claims abstract description 47
- 210000003855 cell nucleus Anatomy 0.000 title claims abstract description 40
- 238000012549 training Methods 0.000 claims abstract description 28
- 238000002372 labelling Methods 0.000 claims abstract description 22
- 230000001575 pathological effect Effects 0.000 claims abstract description 15
- 230000006870 function Effects 0.000 claims description 24
- 230000008569 process Effects 0.000 claims description 19
- 238000005457 optimization Methods 0.000 claims description 13
- 238000004364 calculation method Methods 0.000 claims description 10
- 230000002457 bidirectional effect Effects 0.000 claims description 6
- 238000010586 diagram Methods 0.000 claims description 6
- 206010028980 Neoplasm Diseases 0.000 abstract description 7
- 230000007170 pathology Effects 0.000 abstract description 7
- 238000000605 extraction Methods 0.000 abstract description 6
- 238000012805 post-processing Methods 0.000 abstract description 6
- 238000003759 clinical diagnosis Methods 0.000 abstract description 2
- 238000011160 research Methods 0.000 abstract description 2
- 201000011510 cancer Diseases 0.000 description 5
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 5
- 210000004027 cell Anatomy 0.000 description 4
- 238000004891 communication Methods 0.000 description 4
- 238000010827 pathological analysis Methods 0.000 description 4
- 230000004913 activation Effects 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 210000004940 nucleus Anatomy 0.000 description 2
- 238000004393 prognosis Methods 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 208000030808 Clear cell renal carcinoma Diseases 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 206010073251 clear cell renal cell carcinoma Diseases 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 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
- 238000004043 dyeing Methods 0.000 description 1
- 239000000446 fuel Substances 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000036210 malignancy Effects 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000010186 staining Methods 0.000 description 1
- 230000008093 supporting effect Effects 0.000 description 1
- 238000012360 testing method 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
-
- 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
- G06T2207/10056—Microscopic image
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20152—Watershed segmentation
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a cell nucleus segmentation system and a cell nucleus segmentation method based on label correction, wherein the system comprises a cell nucleus extraction module, a segmentation correction module and a post-processing module; the method comprises the following steps: extracting all connected domains for an original pathological picture and corresponding part noise labeling, and carrying out pixel-level mask correction; the noise label and the corresponding original image of each extracted connected domain are marked, the correction of the noise label is completed through the label correction network, and the training of the segmentation network is supervised; after the correction mask of each connected domain noise label is obtained, the mark of the overlapped cell nucleus is segmented by using a watershed algorithm with an identifier for all correction masks, and finally, the segmentation boundary can be obtained for each cell nucleus. The network model trained by the invention can accurately identify the boundary outline of each cell nucleus, assist the clinical diagnosis of pathology, improve the working efficiency of pathologists, and support the tasks of downstream tumor microenvironment research and the like.
Description
Technical Field
The invention belongs to the technical field of medical image processing and computer vision, and particularly relates to a meta-learning cell nucleus segmentation system and method based on label correction.
Background
Pathological diagnosis is a "gold standard" for cancer screening. The task of cell nucleus segmentation is to separate all cell nuclei from the background from the pathology image. Accurate segmentation of all cell checkups is critical to aid in clinical pathology diagnosis. In traditional pathological diagnosis, a pathologist needs to manually find all cell nuclei under a microscope field of view and judge the malignancy degree of the cell nuclei, then determine the cancer grade through morphological characteristics of the cell nuclei such as nuclear-to-mass ratio, area and the like, give a pathological diagnosis report, and help to promote prognosis of patients. A pathological section usually contains millions of nuclei and the staining of the section is uneven, which presents a great challenge for rapid and accurate pathological diagnosis. In digital pathological image analysis, accurate segmentation of cell nuclei also has important supporting effects on downstream tasks such as genotype phenotype association, prognosis analysis and the like.
Classical machine learning methods are affected by pathological section dyeing quality, cell nucleus polymorphism and the like, so that it is difficult to establish a model with good generalization performance through the characteristics of artificial design, and accurate segmentation of cell nuclei is realized. In recent years, with the development of artificial intelligence techniques typified by deep learning, a deep learning model has achieved good segmentation performance in a plurality of cancer environments. The large scale labeled data is the fuel for training a good deep learning model. The pathology image contains complex phenotypic information such as a large number of nuclei, dense overlap, and usually occurs in clusters or clusters. Labeling of medical images such as pathology images often also has a certain medical knowledge bottleneck. This results in difficult acquisition of large-scale fine-grained fine labeling pathology data, and further results in data labeling often containing some amount of noise due to low confidence among different labels. It is important how to train a model with good segmentation performance using data with partial labels or with noisy labels.
Currently, there are two main types of methods for solving training by using partial labeling or noise labeling. The first category is to introduce additional priori knowledge, artificially generate additional constraint, assist in supervised model training and alleviate the problem of insufficient data annotation to a certain extent. For example, a clustering label and a Thiessen polygon supervision Bayesian network are generated based on point sparse labeling, and cell nuclei with large uncertainty are selected from the clustering label and delivered to a labeling person for labeling, so that the labeling burden is greatly relieved. The main problem of this kind of method is that it needs to generate pseudo tag based on external knowledge and perform iterative optimization, when the data contains more noise, the model is hard to converge to better performance. The second is a method employing loss weighting. The influence of noise labeling loss on model training is reduced through learning a weight matrix. For example, the contribution of each pixel point to model optimization is adjusted by generating importance weights for each pixel loss gradient direction, and the influence of noise labeling during model training is weakened. The limitation of this type of method is that only the weight of the instance contribution in the learning process can be increased or decreased, and there is a problem of information bottleneck, since this method weights the importance of the loss function, if the losses of different input pairs are the same, the learned importance weights cannot be effectively distinguished.
Disclosure of Invention
The invention aims to provide a label correction-based meta-learning cell nucleus segmentation system and a label correction-based meta-learning cell nucleus segmentation method, which are used for overcoming the defects of the prior art.
In order to achieve the above purpose, the invention adopts the following technical scheme:
a meta-learning cell nucleus segmentation system based on label correction comprises a cell nucleus extraction module, a segmentation correction module and a post-processing module;
the cell nucleus extraction module pre-trains a segmentation model according to the partial labeling or the noise labeling to obtain an initial segmentation result, extracts each communication area in the initial segmentation result and inputs the communication area into the next module;
the segmentation correction module trains a segmentation network and a label correction network according to the noise labels and the corresponding pictures under the assistance of five-percent fine granularity precision marking of the total sample number, and completes correction of the noise cell nuclear labels; the method comprises the steps that a segmentation network inputs a pathological picture to generate a segmentation mask feature map, and a label correction network inputs the segmentation feature map and a noise label to generate a correction label to supervise training of the segmentation network;
the post-processing module separates the overlapped cell nucleus masks by adopting a watershed algorithm based on the identifier on the basis of the output result of the segmentation correction module.
A meta-learning cell nucleus segmentation method based on label correction comprises the following steps:
step one:
training a segmentation network by using the part of noise marking data for an original pathological picture and the corresponding part of noise marking, carrying out preliminary prediction on the original pathological picture by the segmentation network, extracting all connected domains according to an initial prediction mask, and then carrying out pixel-level mask correction;
step two:
designing a label correction network in a full convolution form by adopting a meta-learning idea based on label correction for each connected domain noise label and corresponding original image extracted in the step one, completing correction of noise labels through the label correction network under the assistance of five percent label of total sample number, and supervising the training of a segmentation network;
step three:
after the correction masks of the noise labels of each connected domain are obtained, dividing labels of overlapped cell nuclei by using a watershed algorithm with identifiers for all correction masks, and finally obtaining a division boundary for each cell nucleus.
The invention is further improved in that, in the second step,
with D= { x, y } m Representing a small number of clean data samples, D '= { x, y' } M Representing a data sample containing noise annotations; where M, M represents the number of noise and clean samples, y' represents the noise sample signature, the partitioning network is parameterized as a function with parameter W, y=f W (x) The tag correction network is formed as a function of the parameter θ, y c ＝g θ (h (x), y'); wherein h (x) represents a characteristic diagram of the output of the segmentation network, y c Representing the label corrected by the label correction network; training the segmentation network and the label correction network as a bidirectional optimization process; when the label correction network parameter theta is fixed, obtaining an optimal W parameter by minimizing an objective function of the formula (1);
by means of a metadata set containing five percent of total samples and determined optimal parameters of the segmentation network
Optimizing the objective function of the formula (2) to obtain an optimal theta value;
the invention is further improved in that in the actual training process, the values of W and theta are alternately updated by adopting a bidirectional optimization method.
The invention is further improved in that in order to obtain optimal segmentation network and label correction network parameters, the values of W and θ are updated alternately in a single cycle in an iterative manner.
The invention further improves the one-time circulation optimization algorithm process, which comprises the following steps:
1. initializing split network parameters W (0) And tag correction network parameter θ (0) ；
2. In the process of the t iteration, the parameters of the segmentation network are temporarily updated according to a formula (3), and an objective function formula (1) is minimized through one-time gradient update, wherein a calculation formula of the loss at the t moment is different from a calculation formula of the loss at the t moment (4);
3. updating the tag correction network parameter theta by minimizing the objective function (2), wherein the updating process is shown as a formula (5), and a loss calculation formula at the time t+1 is shown as a formula (6);
4. finally, updating the split network parameters W by minimizing the split network objective function (1); the loss calculation formula at the time t+1 is shown as (8);
compared with the prior art, the invention has at least the following beneficial technical effects:
1. based on the idea of label correction, a new meta-learning framework is provided, a label correction network maps a data characteristic diagram and a noise label into correction labels, and the influence of noise labeling is effectively restrained under the granularity of a pixel level, so that the segmentation network training is assisted.
2. The burden of fine-granularity fine labeling data is greatly relieved, a labeling person only needs to add part of labels, the labeled boundaries do not need to be too accurate, and the segmentation effect of the supervision scene can be achieved by using the method.
3. The bottleneck and threshold of pathological section marking are reduced, and a marker can finish the task of data marking without having a professional medical knowledge background like a pathologist.
4. The network model trained by the invention can accurately identify the boundary outline of each cell nucleus, assist the clinical diagnosis of pathology, improve the working efficiency of pathologists, and support the tasks of downstream tumor microenvironment research and the like.
Drawings
Fig. 1 is a block diagram of a segmentation system, comprising three modules: the cell nucleus extraction module, the segmentation correction module and the post-processing module.
Fig. 2 is a specific structure and an optimization process of the segmentation correction module, and mainly includes two parts of a segmentation network and a label correction network.
Fig. 3 is a structure of a split network comprising five convolutional layers and residual connection.
Fig. 4 is a diagram of a tag correction network structure including two convolutional layers and one activation function layer.
Detailed Description
Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art. It should be noted that, without conflict, the embodiments of the present invention and features of the embodiments may be combined with each other. The invention will be described in detail below with reference to the drawings in connection with embodiments.
As shown in FIG. 1, the meta-learning cell nucleus segmentation system based on label correction provided by the invention comprises a cell nucleus extraction module, a segmentation correction module and a post-processing module.
The cell nucleus extraction module pre-trains a segmentation model according to part of labels or noise labels to obtain an initial segmentation result, extracts each communication area in the initial segmentation result and inputs the communication area into the next module; the segmentation correction module trains a segmentation network and a label correction network according to the noise labels and the corresponding pictures with the aid of five-percent fine granularity precision labeling of the total sample number, and completes correction of the noise cell nuclear labels; the method comprises the steps of inputting pathological pictures through a segmentation network to generate a segmentation mask feature map, wherein the segmentation network structure is shown in fig. 3 and mainly comprises a five-layer convolution structure, and feature fusion is carried out between different layers through residual connection. The label correction network inputs the segmentation feature map and the noise label to generate the training of the correction label supervision segmentation network, and the structure of the label correction network is shown in fig. 4, and mainly comprises two convolution layers and a final activation function layer;
the post-processing module separates the overlapped cell nucleus masks by adopting a watershed algorithm based on the identifier on the basis of the output result of the segmentation correction module.
The invention provides a meta-learning cell nucleus segmentation method based on label correction, which comprises the following specific implementation steps:
step one:
training a segmentation network by using the part of noise labeling data for an original pathological picture and the corresponding part of noise labeling, primarily predicting the original pathological picture by the network, extracting all connected domains according to an initial prediction mask, and then placing the extracted connected domains into a segmentation correction module for pixel-level mask correction;
step two:
and (3) for the noise label and the corresponding original image of each connected domain extracted in the step one, designing a label correction network in a full convolution form by adopting a meta-learning idea based on label correction, and completing correction of noise labels and training of a supervision and segmentation network through the label correction network under the assistance of five-percent label of total sample number.
With D= { x, y } m Representing a small number of clean data samples, D '= { x, y' } M Representing a data sample containing noise annotations; where M, M represents the number of noise and clean samples, y' represents the noise sample signature, the partitioning network is parameterized as a function with parameter W, y=f W (x) The tag correction network is formed as a function of the parameter θ, y c ＝g θ (h (x), y'); wherein h (x) represents a characteristic diagram of the output of the segmentation network, y c Representing the label corrected by the label correction network; training the segmentation network and the label correction network as a bidirectional optimization process; when the tag correction network parameter theta is fixed, the optimal W parameter is obtained by minimizing the objective function of the formula (1).
By means of a metadata set containing five percent of total samples and determined optimal parameters of the segmentation network
Optimizing the objective function of the formula (2) to obtain the optimal theta value. />
In the actual training process, the values of W and theta are updated alternately by adopting a bidirectional optimization method.
In order to obtain optimal segmentation network and label correction network parameters, the values of W and theta are alternately updated in a one-time loop process in an iterative manner. The one-time loop optimization algorithm process mainly comprises the following steps:
1. initializing split network parameters W (0) And tag correction network parameter θ (0) 。
2. In the process of the t-th iteration, the parameters of the segmentation network are temporarily updated according to the formula (3), and the objective function formula (1) is minimized through one-time gradient update. The calculation formula of the loss at the time t is shown as (4).
3. The tag correction network parameter θ can be updated by minimizing the objective function (2), the update procedure being as in equation (5). The loss calculation formula at time t+1 is shown in (6).
4. Finally, updating the split network parameters W is accomplished by minimizing the split network objective function (1).
The loss calculation formula at time t+1 is shown in (8).
The one-time loop overall optimization flow is shown in fig. 2, (1) inputting an image into a current segmentation network, then calculating a logical stoneley graph of a prediction result, (2) inputting a part of noise mask and the logical stoneley graph of the prediction result into a label correction network to obtain a correction mask, (3) calculating a loss gradient of the logical stoneley graph and the correction mask, then calculating a loss gradient related to parameters of the segmentation network, (4) updating the parameters of the segmentation network while maintaining the gradient map, (5) inputting a pair of masks and pictures marked with fine granularity into a new segmentation network, and calculating the loss, (6) calculating the loss gradient, and updating the parameters of the label correction network.
Step three:
after the correction masks of all noise labels are obtained, a watershed algorithm with identifiers is adopted for all labels, labels of overlapped cell nuclei are segmented, and finally good segmentation boundaries can be obtained for each cell nucleus.
Examples
The present embodiment combines the above proposed method to segment cancer cells in pathological sections of renal clear cell carcinoma. Comprising the following steps:
(1) And generating a data set. And cutting the image blocks and the mask blocks with the sizes of 64 times 64 according to the segmentation mask of the pathological section to be used as a construction data set. The number of training sets is 2000, the number of test sets is 1500, and the number of metadata sets is 50.
(2) And generating a noise mask. In the embodiment, two noise generation modes are adopted, wherein the first mode is that part of gold standard marks are marked, 40% of gold standard masks are randomly selected for reservation, and the rest mark masks are deleted; the second is a part of weak label, and the gold standard is expanded or expanded into a square frame by randomly expanding one to three pixels on the basis of the part of gold standard label.
(3) Implementation details. The embodiment is realized by adopting a Pytorch framework, the segmentation network adopts a Resnet-32 structure as a feature extractor, and the network structure adopts a U-shaped structure. The tag correction network uses a layer 3 by 3 convolution and a layer 1 by 1 convolution. During training, the learning rate of the segmentation network is set to be 1x10 -3 Drop 0.1 after 300 generation training, tag calibrationThe learning rate of the positive network is set to 1x10 -4 Adam was used as an optimizer for both networks.
(4) And outputting a result. After training is completed, the original image is input into the segmentation network to output the segmentation result of the cell nucleus, the label correction network only assists the segmentation model training in the training process, the experimental result is shown in table 1, and the proposed method even achieves the performance of a supervision scene under the conditions of expansion noise and part of gold standard labeling scenes. Compared with the method for training by directly using noise data, the method has the advantage that the performance is greatly improved under three noise scenes.
Table 1 shows experimental results of the proposed segmentation method used in the examples.
While the invention has been described in detail in the foregoing general description and with reference to specific embodiments thereof, it will be apparent to one skilled in the art that modifications and improvements can be made thereto. Accordingly, such modifications or improvements may be made without departing from the spirit of the invention and are intended to be within the scope of the invention as claimed.
Claims (1)
1. The meta-learning cell nucleus segmentation method based on label correction is characterized by comprising the following steps of:
step one:
training a segmentation network by using the part of noise marking data for an original pathological picture and the corresponding part of noise marking, carrying out preliminary prediction on the original pathological picture by the segmentation network, extracting all connected domains according to an initial prediction mask, and then carrying out pixel-level mask correction;
step two:
for each connected domain noise label and corresponding original image extracted in the step one, a label correction network is designed in a full convolution form by adopting a meta learning idea based on label correction, and the number of total samples is calculated by using the methodWith the aid of five percent labeling, the correction of the noise label is completed through the label correction network, and the training of the segmentation network is supervised; with D= { x, y } m Represent a small number of clean data samples, D ′ ＝{x,y ′ } M Representing a data sample containing noise annotations; wherein M, M represents the number of noise and clean samples, y ′ Representing noise sample labels, the split network parameterizes as a function with parameter W, y=f W (x) The tag correction network is formed as a function of the parameter θ, y c ＝g θ (h(x),y ′ ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein h (x) represents a characteristic diagram of the output of the segmentation network, y c Representing the label corrected by the label correction network; training the segmentation network and the label correction network as a bidirectional optimization process; when the label correction network parameter theta is fixed, obtaining an optimal W parameter by minimizing an objective function of the formula (1);
by means of a metadata set containing five percent of total samples and determined optimal parameters of the segmentation network
Optimizing the objective function of the formula (2) to obtain an optimal theta value;
in the actual training process, the values of W and theta are updated alternately by adopting a bidirectional optimization method;
in order to obtain optimal parameters of the segmentation network and the label correction network, adopting an iterative mode, and alternately updating the values of W and theta in a one-time circulation process;
a one-time loop optimization algorithm process comprising the steps of:
1) Initializing split network parameters W (0) And tag correction network parameter θ (0) ；
2) In the process of the t iteration, the parameters of the segmentation network are temporarily updated according to a formula (3), and an objective function formula (1) is minimized through primary gradient update, wherein a calculation formula of the loss at the t moment is shown as (4);
3) Updating the tag correction network parameter theta by minimizing the objective function (2), wherein the updating process is shown as a formula (5), and a loss calculation formula at the time t+1 is shown as a formula (6);
4) Finally, updating the split network parameters W by minimizing the split network objective function (1); the loss calculation formula at the time t+1 is shown as (8);
step three:
after the correction masks of the noise labels of each connected domain are obtained, dividing labels of overlapped cell nuclei by using a watershed algorithm with identifiers for all correction masks, and finally obtaining a division boundary for each cell nucleus.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110651067.1A CN113450363B (en) | 2021-06-10 | 2021-06-10 | Meta-learning cell nucleus segmentation system and method based on label correction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110651067.1A CN113450363B (en) | 2021-06-10 | 2021-06-10 | Meta-learning cell nucleus segmentation system and method based on label correction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113450363A CN113450363A (en) | 2021-09-28 |
| CN113450363B true CN113450363B (en) | 2023-05-02 |
Family
ID=77811177
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110651067.1A Active CN113450363B (en) | 2021-06-10 | 2021-06-10 | Meta-learning cell nucleus segmentation system and method based on label correction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113450363B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113962999B (en) * | 2021-10-19 | 2024-06-25 | 浙江大学 | Noise label segmentation method based on Gaussian mixture model and label correction model |
| CN114511552A (en) * | 2022-02-24 | 2022-05-17 | 中南大学湘雅医院 | Thyroid ultrasound nodule fuzzy boundary-oriented segmentation method based on meta-learning |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112102323A (en) * | 2020-09-17 | 2020-12-18 | 陕西师范大学 | Adherent nucleus segmentation method based on generation of countermeasure network and Caps-Unet network |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015089434A1 (en) * | 2013-12-12 | 2015-06-18 | University Of Pittsburgh - Of The Commonwealth System Of Higher Education | Automated epithelial nuclei segmentation for computational disease detection algorithms |
| US10789451B2 (en) * | 2017-11-16 | 2020-09-29 | Global Life Sciences Solutions Usa Llc | System and method for single channel whole cell segmentation |
| CN109493330A (en) * | 2018-11-06 | 2019-03-19 | 电子科技大学 | A kind of nucleus example dividing method based on multi-task learning |
| CN111091571B (en) * | 2019-12-12 | 2024-05-14 | 珠海圣美生物诊断技术有限公司 | Cell nucleus segmentation method, device, electronic equipment and computer readable storage medium |
| CN112132843B (en) * | 2020-09-30 | 2023-05-19 | 福建师范大学 | Hematoxylin-eosin staining pathological image segmentation method based on unsupervised deep learning |
-
2021
- 2021-06-10 CN CN202110651067.1A patent/CN113450363B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112102323A (en) * | 2020-09-17 | 2020-12-18 | 陕西师范大学 | Adherent nucleus segmentation method based on generation of countermeasure network and Caps-Unet network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113450363A (en) | 2021-09-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111401480B (en) | Novel mammary gland MRI automatic auxiliary diagnosis method based on fusion attention mechanism | |
| CN110889852B (en) | Liver segmentation method based on residual error-attention deep neural network | |
| WO2021203795A1 (en) | Pancreas ct automatic segmentation method based on saliency dense connection expansion convolutional network | |
| CN113450363B (en) | Meta-learning cell nucleus segmentation system and method based on label correction | |
| CN111242233B (en) | Alzheimer disease classification method based on fusion network | |
| CN113111916B (en) | Medical image semantic segmentation method and system based on weak supervision | |
| CN110059697A (en) | A kind of Lung neoplasm automatic division method based on deep learning | |
| CN109102498B (en) | Method for segmenting cluster type cell nucleus in cervical smear image | |
| CN108305253A (en) | A kind of pathology full slice diagnostic method based on more multiplying power deep learnings | |
| CN111784564B (en) | Automatic image matting method and system | |
| CN111882620A (en) | Road drivable area segmentation method based on multi-scale information | |
| Chen et al. | Binarized neural architecture search | |
| WO2024108522A1 (en) | Multi-modal brain tumor image segmentation method based on self-supervised learning | |
| CN112837338A (en) | Semi-supervised medical image segmentation method based on generation countermeasure network | |
| CN117437423A (en) | Weak supervision medical image segmentation method and device based on SAM collaborative learning and cross-layer feature aggregation enhancement | |
| CN115601330A (en) | Colonic polyp segmentation method based on multi-scale space reverse attention mechanism | |
| CN116363149A (en) | Medical image segmentation method based on U-Net improvement | |
| CN112508860B (en) | Artificial intelligence interpretation method and system for positive check of immunohistochemical image | |
| CN113012167B (en) | Combined segmentation method for cell nucleus and cytoplasm | |
| CN117151162A (en) | Cross-anatomical-area organ incremental segmentation method based on self-supervision and specialized control | |
| CN115578260B (en) | Attention method and system for directional decoupling of image super-resolution | |
| CN116778164A (en) | Semantic segmentation method for improving deep V < 3+ > network based on multi-scale structure | |
| CN113888551A (en) | Liver tumor image segmentation method based on dense connection network of high-low layer feature fusion | |
| CN116258660A (en) | Intelligent measuring system for invasion depth of oral squamous carcinoma | |
| CN110969628B (en) | Super-pixel segmentation method based on variation level set |
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 |