CN113034448A - Pathological image cell identification method based on multi-instance learning - Google Patents
Pathological image cell identification method based on multi-instance learning Download PDFInfo
- Publication number
- CN113034448A CN113034448A CN202110263177.0A CN202110263177A CN113034448A CN 113034448 A CN113034448 A CN 113034448A CN 202110263177 A CN202110263177 A CN 202110263177A CN 113034448 A CN113034448 A CN 113034448A
- Authority
- CN
- China
- Prior art keywords
- cell
- image
- tissue
- learning
- method based
- 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
- 230000001575 pathological effect Effects 0.000 title claims abstract description 23
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000007781 pre-processing Methods 0.000 claims abstract description 4
- 238000012549 training Methods 0.000 claims description 21
- 230000006870 function Effects 0.000 claims description 13
- 238000012360 testing method Methods 0.000 claims description 10
- 230000002776 aggregation Effects 0.000 claims description 9
- 238000004220 aggregation Methods 0.000 claims description 9
- 238000005457 optimization Methods 0.000 claims description 7
- 238000013526 transfer learning Methods 0.000 claims description 6
- 230000005859 cell recognition Effects 0.000 claims description 5
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000013508 migration Methods 0.000 claims description 3
- 230000005012 migration Effects 0.000 claims description 3
- 230000007547 defect Effects 0.000 abstract description 8
- 238000013528 artificial neural network Methods 0.000 abstract description 5
- 238000003745 diagnosis Methods 0.000 abstract description 5
- 230000009286 beneficial effect Effects 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000007635 classification algorithm Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000012634 fragment Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000010186 staining Methods 0.000 description 1
- 238000012795 verification 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
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- 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
- G06T2207/20—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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30024—Cell structures in vitro; Tissue sections in vitro
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Radiology & Medical Imaging (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medical Informatics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Investigating Or Analysing Biological Materials (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a pathological image cell identification method based on multi-instance learning, which comprises the following steps of: s1: collecting a cell tissue image and preprocessing the cell tissue image; s2: establishing a prediction model; s3: and inputting the cell image data to be detected into the prediction model to perform pathological image cell identification. The invention trains the single cell picture by using the label of the whole tissue picture, so that the network has the capacity of identifying the single cell, and the defect that the traditional neural network trained by the whole picture cannot predict the single cell is overcome. The invention makes up the defect that the traditional prediction model can not predict the cells without labels by using a multi-example learning method. The model can provide an important technical means for constructing an automatic auxiliary diagnosis and treatment system in medical intelligent application.
Description
Technical Field
The invention belongs to the technical field of machine learning and deep learning, and particularly relates to a pathological image cell identification method based on multi-instance learning.
Background
The pathological image is a section image obtained by H & E staining of a diseased tissue taken out of the body of a subject. Since this dye only reacts with cells, only cells in the tissue structure stained by H & E will be stained with color. So that the pathologist can directly observe and count the cells. It has important effect on the diagnosis of the diseased condition of the patient. The traditional detection method is that doctors directly use a microscope to observe or visually observe an enlarged picture to identify cells, find positive cells and calculate the positive rate. The manual identification has the advantages of high accuracy and no error. But the defects are also obvious, different people may judge differently on certain cells, and the judgment is subjective; time spent by trained professionals and inefficiency; often the observations are made by estimation. For these reasons, a large number of pathological images are either unlabeled, or the entire pathological image has only "cancerous" and "non-cancerous" labels, while specific regions and specific cells in the picture are unlabeled.
There are many ways of processing such pictures by computers. In the traditional picture processing method, some operators are used to obtain features, and then the features are subjected to statistical operation. Or the image is segmented manually, the region with the characteristics is calculated to obtain the characteristic vector, and finally the classifier is used for operation. However, since the data with cell labels is too small, it is difficult to train a classifier for the cells themselves.
Disclosure of Invention
The invention aims to solve the problem of a general classification algorithm and provides a pathological image cell identification method based on multi-example learning.
The technical scheme of the invention is as follows: a pathological image cell identification method based on multi-instance learning comprises the following steps:
s1: collecting a cell tissue image and preprocessing the cell tissue image;
s2: establishing a prediction model based on the preprocessed image;
s3: and inputting the cell image data to be detected into the prediction model to perform pathological image cell identification.
The invention has the beneficial effects that: the invention utilizes the labels of the single cell picture and the whole tissue picture to train, so that the network has the capability of identifying the single cell and makes up the defect that the traditional neural network trained by the whole picture can not predict the single cell. The invention makes up the defect that the traditional prediction model can not predict the cells without labels by using a multi-example learning method. The model can provide an important technical means for constructing an automatic auxiliary diagnosis and treatment system in medical intelligent application.
Further, step S1 includes the following sub-steps:
s11: collecting a cell tissue image, and storing cell marks in the cell tissue image in a CSV file form;
s12: taking the column of the CSV file as the abscissa and the ordinate of the central point of the cell tissue, and dividing the column into a training set and a testing set;
s13: randomly selecting one of the cell tissue images, and training a mean value device by using a stationools bag;
s14: and normalizing the training set and the test set by using the trained averager, and reading the normalized cell image by using the sci-image based on the abscissa and the ordinate of the central point of the cell tissue to finish the pretreatment.
The beneficial effects of the further scheme are as follows: in the invention, all the picture files of the training set and the test set are normalized by a mean value device Normalizer and then are calculated. The resulting image contains both the features of the cell and the features surrounding the cell.
Further, step S2 includes the following sub-steps:
s21: taking the ResNet network as a feature extractor, outputting the features of the single cells, and performing transfer learning;
s22: taking each cell feature after migration learning in the normalized whole cell tissue image as a positive rate feature of the image, and inputting the cell features into the multi-example learning layer for aggregation;
s23: and taking the output after aggregation as the prediction probability of the whole cell tissue image, and training by utilizing an optimization function to complete the establishment of a prediction model.
The beneficial effects of the further scheme are as follows: in the present invention, a high-performance ResNet network is used in order to extract cell features efficiently and quickly. On the other hand, since the cell image is small, a relatively small network such as ResNet50 or ResNet18 in the ResNet family is generally used.
Further, in step S21, if the number of cell tissue images is less than 200, a ResNet18 network is used; if the number of the cell tissue images is more than 200, adopting a ResNet50 network; when the transfer learning is carried out, a linear layer with the input shape of 1000 multiplied by 1 and the output of the linear layer with the corresponding classification number is added to the output layer of the ResNet network.
Further, in step S22, the calculation formula for the multi-instance learning layer to aggregate is:
wherein, g ({ p)j}) represents the positive probability of the whole tissue image, | j | represents the number of cells contained in each tissue cell image, pjThe probability of positive prediction for each individual cell is indicated, and r is the hyperparameter.
Further, in step S23, the optimization function is a cross entropy loss function, the optimizer is set as Adam optimizer, and the learning rate is 0.001.
Drawings
Fig. 1 is a flowchart of a pathological image cell identification method.
Detailed Description
The embodiments of the present invention will be further described with reference to the accompanying drawings.
Before describing specific embodiments of the present invention, in order to make the solution of the present invention more clear and complete, the definitions of the abbreviations and key terms appearing in the present invention will be explained first:
sci-image: an image processing software package.
As shown in fig. 1, the invention provides a pathological image cell recognition method based on multi-instance learning, comprising the following steps:
s1: collecting a cell tissue image and preprocessing the cell tissue image;
s2: establishing a prediction model based on the preprocessed image;
s3: and inputting the cell image data to be detected into the prediction model to perform pathological image cell identification.
In the embodiment of the present invention, as shown in fig. 1, step S1 includes the following sub-steps:
s11: collecting a cell tissue image, and storing cell marks in the cell tissue image in a CSV file form;
s12: taking the column of the CSV file as the abscissa and the ordinate of the central point of the cell tissue, and dividing the column into a training set and a testing set;
s13: randomly selecting one of the cell tissue images, and training a mean value device by using a stationools bag;
s14: and normalizing the training set and the test set by using the trained averager, and reading the normalized cell image by using the sci-image based on the abscissa and the ordinate of the central point of the cell tissue to finish the pretreatment.
In the invention, all the picture files of the training set and the test set are normalized by a mean value device Normalizer and then are calculated. The resulting image contains both the features of the cell and the features surrounding the cell. The test and training sets are typically 7: 3 or 8: the ratio of 2 was chosen randomly from the data samples.
In the embodiment of the present invention, as shown in fig. 1, step S2 includes the following sub-steps:
s21: taking the ResNet network as a feature extractor, outputting the features of the single cells, and performing transfer learning;
s22: taking each cell feature after migration learning in the normalized whole cell tissue image as a positive rate feature of the image, and inputting the cell features into the multi-example learning layer for aggregation;
s23: and taking the output after aggregation as the prediction probability of the whole cell tissue image, and training by utilizing an optimization function to complete the establishment of a prediction model.
In the present invention, a high-performance ResNet network is used in order to extract cell features efficiently and quickly. On the other hand, since the cell image is small, a relatively small network such as ResNet50 or ResNet18 in the ResNet family is generally used.
In the embodiment of the present invention, as shown in fig. 1, in step S21, if the number of cell tissue images is less than 200, a ResNet18 network is adopted; if the number of the cell tissue images is more than 200, adopting a ResNet50 network; when the transfer learning is carried out, a linear layer with the input shape of 1000 multiplied by 1 and the output of the linear layer with the corresponding classification number is added to the output layer of the ResNet network.
In the embodiment of the present invention, as shown in fig. 1, in step S22, the calculation formula of aggregation performed by the multiple example learning layers is as follows:
wherein, g ({ p)j}) represents the positive probability of the whole tissue image, | j | represents the number of cells contained in each tissue cell image, pjThe probability of positive prediction for each individual cell is indicated, and r is the hyperparameter.
In the embodiment of the present invention, as shown in fig. 1, in step S23, the optimization function is a cross entropy loss function, the optimizer is set as Adam optimizer, and the learning rate is 0.001.
In the embodiment of the invention, the prediction model is generated by automatic training of an algorithm framework by using a loss function provided by various deep learning libraries (such as a pytorch, tenserflow and the like) and an interface of a computation layer. The realization process is as follows: programming to realize a self-defined neural network interface according to the content of the step S21; programming a multi-instance learning layer as per the contents of step S22; setting model parameters, and training to obtain a prediction model by using data defined in the S1 stage and used for establishing the prediction model; and verifying the model through the test set and the verification set, debugging the hyperparameter T, obtaining an ROC curve and calculating an AUC value.
The working principle and the process of the invention are as follows: the invention discloses a method for predicting positive cells in a medical pathological tissue image. The method is based on a deep neural network model, utilizes an aggregation function thought in a multi-example learning algorithm, and utilizes image-level clinical label information to construct automatic image classification and positive cell counting based on positive cell estimation. The invention is divided into two stages of training and forecasting. In the training stage, a ResNet pre-training network of a deep neural network is adopted, a Pythrch frame is used, the feature expression of the fragment image data taking the cell as the center is extracted, and the feature is used as the single cell feature depiction. The positive probability of the predicted cell image is characterized by the characteristics, and an image-level multi-example learning optimization function is constructed by using a multi-example learning aggregation function, so that a prediction model is obtained through training. In the prediction stage, a pathological tissue image is input, a cell detection algorithm is used for detecting and cutting out a cell image packet, cell image characteristics are extracted, a multi-example learning idea is utilized, the tissue image is classified and predicted, and the positive cell probability is output. The invention applies a multi-example learning method to make up the defect that the traditional prediction model can not use the image-level label to carry out cell classification prediction. The model can provide an important technical means for constructing an automatic auxiliary diagnosis and treatment system in medical intelligent application.
The invention has the beneficial effects that: the invention utilizes the labels of the single cell picture and the whole tissue picture to train, so that the network has the capability of identifying the single cell and makes up the defect that the traditional neural network trained by the whole picture can not predict the single cell. The invention makes up the defect that the traditional prediction model can not predict the cells without labels by using a multi-example learning method. The model can provide an important technical means for constructing an automatic auxiliary diagnosis and treatment system in medical intelligent application.
It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Those skilled in the art can make various other specific changes and combinations based on the teachings of the present invention without departing from the spirit of the invention, and these changes and combinations are within the scope of the invention.
Claims (6)
1. A pathological image cell identification method based on multi-instance learning is characterized by comprising the following steps:
s1: collecting a cell tissue image and preprocessing the cell tissue image;
s2: establishing a prediction model based on the preprocessed image;
s3: and inputting the cell image data to be detected into the prediction model to perform pathological image cell identification.
2. The pathological image cell recognition method based on multi-instance learning according to claim 1, wherein the step S1 includes the following sub-steps:
s11: collecting a cell tissue image, and storing cell marks in the cell tissue image in a CSV file form;
s12: taking the column of the CSV file as the abscissa and the ordinate of the central point of the cell tissue, and dividing the column into a training set and a testing set;
s13: randomly selecting one of the cell tissue images, and training a mean value device by using a stationools bag;
s14: and normalizing the training set and the test set by using the trained averager, and reading the normalized cell image by using the sci-image based on the abscissa and the ordinate of the central point of the cell tissue to finish the pretreatment.
3. The pathological image cell recognition method based on multi-instance learning according to claim 1, wherein the step S2 includes the following sub-steps:
s21: taking the ResNet network as a feature extractor, outputting the features of the single cells, and performing transfer learning;
s22: taking each cell feature after migration learning in the normalized whole cell tissue image as a positive rate feature of the image, and inputting the cell features into the multi-example learning layer for aggregation;
s23: and taking the output after aggregation as the prediction probability of the whole cell tissue image, and training by utilizing an optimization function to complete the establishment of a prediction model.
4. The pathological image cell recognition method based on multi-instance learning according to claim 3, wherein in step S21, if the number of cell tissue images is less than 200, a ResNet18 network is adopted; if the number of the cell tissue images is more than 200, adopting a ResNet50 network; when the transfer learning is carried out, a linear layer with the input shape of 1000 multiplied by 1 and the output of the linear layer with the corresponding classification number is added to the output layer of the ResNet network.
5. The pathological image cell recognition method based on multi-instance learning according to claim 3, wherein in step S22, the calculation formula for the multi-instance learning layer to aggregate is:
wherein, g ({ p)j}) represents the positive probability of the whole tissue image, | j | represents the number of cells contained in each tissue cell image, pjThe probability of positive prediction for each individual cell is indicated, and r is the hyperparameter.
6. The pathological image cell identification method based on multi-instance learning according to claim 3, wherein in step S23, the optimization function is a cross entropy loss function, the optimizer is set as Adam optimizer, and the learning rate is 0.001.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110263177.0A CN113034448B (en) | 2021-03-11 | 2021-03-11 | Pathological image cell identification method based on multi-instance learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110263177.0A CN113034448B (en) | 2021-03-11 | 2021-03-11 | Pathological image cell identification method based on multi-instance learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113034448A true CN113034448A (en) | 2021-06-25 |
| CN113034448B CN113034448B (en) | 2022-06-21 |
Family
ID=76469475
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110263177.0A Expired - Fee Related CN113034448B (en) | 2021-03-11 | 2021-03-11 | Pathological image cell identification method based on multi-instance learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113034448B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116485792A (en) * | 2023-06-16 | 2023-07-25 | 中南大学 | Histopathological subtype prediction method and imaging method |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050263599A1 (en) * | 2003-11-13 | 2005-12-01 | Metrologic Instruments, Inc. | Digital imaging-based bar code symbol reading system employing a multi-mode image-processing symbol reading subsystem that switches its modes of reading during a single bar code symbol reading cycle, and within each said mode of reading, automatically applies a different image-processing based bar code symbol reading methodology |
| CN108346145A (en) * | 2018-01-31 | 2018-07-31 | 浙江大学 | The recognition methods of unconventional cell in a kind of pathological section |
| CN108717554A (en) * | 2018-05-22 | 2018-10-30 | 复旦大学附属肿瘤医院 | A kind of thyroid tumors histopathologic slide image classification method and its device |
| CN110211108A (en) * | 2019-05-29 | 2019-09-06 | 武汉兰丁医学高科技有限公司 | A kind of novel abnormal cervical cells automatic identifying method based on Feulgen colouring method |
| US20190295252A1 (en) * | 2018-03-23 | 2019-09-26 | Memorial Sloan Kettering Cancer Center | Systems and methods for multiple instance learning for classification and localization in biomedical imaging |
| CN110491505A (en) * | 2019-08-22 | 2019-11-22 | 北京大学第三医院（北京大学第三临床医学院） | The system for predicting the egg mother cell quantity obtained during subject's ovarian stimulation |
| CN111860591A (en) * | 2020-06-16 | 2020-10-30 | 南通大学 | Cervical cell image classification method based on interval adaptive feature selection fusion |
| CN112233061A (en) * | 2020-09-07 | 2021-01-15 | 中国人民解放军海军军医大学第一附属医院 | Deep learning-based skin basal cell carcinoma and Babylonia disease identification method |
| CN112419306A (en) * | 2020-12-11 | 2021-02-26 | 长春工业大学 | Lung nodule detection method based on NAS-FPN |
-
2021
- 2021-03-11 CN CN202110263177.0A patent/CN113034448B/en not_active Expired - Fee Related
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050263599A1 (en) * | 2003-11-13 | 2005-12-01 | Metrologic Instruments, Inc. | Digital imaging-based bar code symbol reading system employing a multi-mode image-processing symbol reading subsystem that switches its modes of reading during a single bar code symbol reading cycle, and within each said mode of reading, automatically applies a different image-processing based bar code symbol reading methodology |
| CN108346145A (en) * | 2018-01-31 | 2018-07-31 | 浙江大学 | The recognition methods of unconventional cell in a kind of pathological section |
| US20190295252A1 (en) * | 2018-03-23 | 2019-09-26 | Memorial Sloan Kettering Cancer Center | Systems and methods for multiple instance learning for classification and localization in biomedical imaging |
| CN108717554A (en) * | 2018-05-22 | 2018-10-30 | 复旦大学附属肿瘤医院 | A kind of thyroid tumors histopathologic slide image classification method and its device |
| CN110211108A (en) * | 2019-05-29 | 2019-09-06 | 武汉兰丁医学高科技有限公司 | A kind of novel abnormal cervical cells automatic identifying method based on Feulgen colouring method |
| CN110491505A (en) * | 2019-08-22 | 2019-11-22 | 北京大学第三医院（北京大学第三临床医学院） | The system for predicting the egg mother cell quantity obtained during subject's ovarian stimulation |
| CN111860591A (en) * | 2020-06-16 | 2020-10-30 | 南通大学 | Cervical cell image classification method based on interval adaptive feature selection fusion |
| CN112233061A (en) * | 2020-09-07 | 2021-01-15 | 中国人民解放军海军军医大学第一附属医院 | Deep learning-based skin basal cell carcinoma and Babylonia disease identification method |
| CN112419306A (en) * | 2020-12-11 | 2021-02-26 | 长春工业大学 | Lung nodule detection method based on NAS-FPN |
Non-Patent Citations (2)
| Title |
|---|
| SONER KAYA 等: "Textural feature extraction and multiple instance learning for cell segmantation in histopathological images", 《2019 SCIENTIFIC MEETING ON ELECTRICAL-ELECTRONICS & BIOMEDICAL ENGINEERING AND COMPUTER SCIENCE (EBBT)》 * |
| 沈晓涵 等: "人工智能在病理诊断领域中的应用", 《肿瘤防治研究》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116485792A (en) * | 2023-06-16 | 2023-07-25 | 中南大学 | Histopathological subtype prediction method and imaging method |
| CN116485792B (en) * | 2023-06-16 | 2023-09-15 | 中南大学 | Histopathological subtype prediction method and imaging method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113034448B (en) | 2022-06-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8379961B2 (en) | Mitotic figure detector and counter system and method for detecting and counting mitotic figures | |
| CN106845421B (en) | Face feature recognition method and system based on multi-region feature and metric learning | |
| CN105095866B (en) | A kind of quick Activity recognition method and system | |
| CN112529042B (en) | Medical image classification method based on dual-attention multi-example deep learning | |
| CN109544518B (en) | Method and system applied to bone maturity assessment | |
| CN109784325A (en) | Opener recognition methods and equipment and computer readable storage medium | |
| CN113256636B (en) | Bottom-up parasite species development stage and image pixel classification method | |
| WO2021051875A1 (en) | Cell classification method and apparatus, medium and electronic device | |
| CN104484886B (en) | A kind of dividing method and device of MR images | |
| CN113628199B (en) | Pathological picture stained tissue area detection method, pathological picture stained tissue area detection system and prognosis state analysis system | |
| CN113724842B (en) | Cervical tissue pathology auxiliary diagnosis method based on attention mechanism | |
| CN115909006A (en) | Mammary tissue image classification method and system based on convolution Transformer | |
| CN113034448B (en) | Pathological image cell identification method based on multi-instance learning | |
| CN112419396A (en) | Thyroid ultrasonic video automatic analysis method and system | |
| CN112579808A (en) | Data annotation processing method, device and system | |
| CN117036288A (en) | Tumor subtype diagnosis method for full-slice pathological image | |
| CN116309459A (en) | Improved network-based lung nodule detection method, apparatus, device and storage medium | |
| CN115206495A (en) | Renal cancer pathological image analysis method and system based on CoAtNet deep learning and intelligent microscopic device | |
| CN116485792A (en) | Histopathological subtype prediction method and imaging method | |
| Ardelean et al. | High-fidelity zero-shot texture anomaly localization using feature correspondence analysis | |
| CN115760707A (en) | Skin damage image intelligent classification device based on self-supervision learning | |
| CN109767430A (en) | The quality determining method and quality detecting system of valuable bills | |
| CN114332007B (en) | Industrial defect detection and identification method based on transducer | |
| CN116188879B (en) | Image classification and image classification model training method, device, equipment and medium | |
| CN113496219B (en) | Automatic blood relationship identification method and device based on face image analysis |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20220621 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |