CN112819765A - Liver image processing method - Google Patents
Liver image processing method Download PDFInfo
- Publication number
- CN112819765A CN112819765A CN202110096604.0A CN202110096604A CN112819765A CN 112819765 A CN112819765 A CN 112819765A CN 202110096604 A CN202110096604 A CN 202110096604A CN 112819765 A CN112819765 A CN 112819765A
- Authority
- CN
- China
- Prior art keywords
- liver
- training
- sample
- distribution
- image processing
- 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.)
- Pending
Links
- 210000004185 liver Anatomy 0.000 title claims abstract description 30
- 238000003672 processing method Methods 0.000 title claims abstract description 15
- 238000012549 training Methods 0.000 claims abstract description 42
- 238000012360 testing method Methods 0.000 claims abstract description 29
- 230000009467 reduction Effects 0.000 claims abstract description 21
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 18
- 238000000034 method Methods 0.000 claims abstract description 10
- 238000007635 classification algorithm Methods 0.000 claims abstract description 9
- 238000009826 distribution Methods 0.000 claims description 30
- 206010019663 Hepatic failure Diseases 0.000 claims description 11
- 208000007903 liver failure Diseases 0.000 claims description 10
- 231100000835 liver failure Toxicity 0.000 claims description 10
- 238000013507 mapping Methods 0.000 claims description 8
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 abstract description 7
- 238000012545 processing Methods 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 11
- 201000007270 liver cancer Diseases 0.000 description 6
- 208000014018 liver neoplasm Diseases 0.000 description 6
- 238000000605 extraction Methods 0.000 description 4
- 206010028980 Neoplasm Diseases 0.000 description 3
- 230000003908 liver function Effects 0.000 description 3
- 238000004393 prognosis Methods 0.000 description 3
- 238000012752 Hepatectomy Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 230000001575 pathological effect Effects 0.000 description 2
- 230000002980 postoperative effect Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000000717 retained effect Effects 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 210000002966 serum Anatomy 0.000 description 2
- 208000000857 Hepatic Insufficiency Diseases 0.000 description 1
- 206010019695 Hepatic neoplasm Diseases 0.000 description 1
- 208000019425 cirrhosis of liver Diseases 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000002440 hepatic effect Effects 0.000 description 1
- 206010073071 hepatocellular carcinoma Diseases 0.000 description 1
- 231100000844 hepatocellular carcinoma Toxicity 0.000 description 1
- 230000036210 malignancy Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000012113 quantitative test Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000001356 surgical procedure Methods 0.000 description 1
- 230000004083 survival effect Effects 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/211—Selection of the most significant subset of features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
- G06T2207/30056—Liver; Hepatic
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Physics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Quality & Reliability (AREA)
- Radiology & Medical Imaging (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Medical Informatics (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- General Health & Medical Sciences (AREA)
- Algebra (AREA)
- Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to the technical field of image processing, in particular to a liver image processing method, which comprises the following steps: firstly, extracting the characteristic of the image group from the CECT image of the liver part through A.K. software; setting a threshold, and removing features lower than the threshold by using a VarianceThreshold method; thirdly, performing feature dimension reduction by using a T-SNE dimension reduction algorithm; fourthly, training and testing the training set samples by utilizing a QN-S3VM semi-supervised classification algorithm; and fifthly, randomly distributing a test set and a training set for multiple times, and taking the average value of the classification accuracy of the test set and the training set. The invention converts the liver image into quantitative and high-dimensional data characteristics, and can assist doctors in liver reserve function evaluation.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to a liver image processing method.
Background
Hepatocellular carcinoma (liver cancer) is one of the most common malignancies worldwide. Hepatectomy is the main means by which liver cancer patients acquire long-term survival. Liver failure is an important complication after hepatectomy, especially hemihepatectomy. Therefore, it is very important to evaluate the liver before operation to avoid the occurrence of postoperative hepatic failure as much as possible, and the accurate evaluation of the hepatic reserve function of the liver cancer patient before operation is not only beneficial to avoiding the hepatic failure and reducing the risk of operation, but also beneficial to reducing death caused by postoperative middle-and long-term hepatic insufficiency. Currently, clinical evaluation means of liver reserve function mainly include liver serum biochemical detection, Child-Pugh score, albumin-bilirubin (ALBI) score, liver function quantitative test, liver volume measurement and the like. The biochemical detection of serum cannot completely reflect liver functions, and the Child-Pugh score cannot completely reflect the prognosis of liver cancer patients. In view of the above, there is still a need to find an effective means for predicting whether or not liver failure will occur after surgery.
The image omics is a new subject direction based on quantitative imaging, and converts medical image data into quantitative and high-dimensional data characteristics through high-flux extraction of quantitative characteristics. The morphological and pathological characteristics of the tumor in the clinical image are described intuitively and quantitatively by utilizing a plurality of image characteristics, and the method has stronger prediction effect on treatment results, prognosis and tumor genetic factors. Research has shown that the characteristics of the image group are related to the pathological characteristics after the liver fibrosis and the liver cancer operation, and the non-tumor liver parenchyma of the CECT image before the operation is presumed to reflect the liver reserve function of the patient.
At present, most of evaluation means of liver reserve function depend on preoperative clinical physiological indexes, and the prognosis condition of liver cancer patients is difficult to be completely reflected. The single dependence on clinical physiological indexes and the high data accuracy of images are difficult to guarantee, so that the interpretation of semantic information is not comprehensive, and the health condition of a testee cannot be accurately judged usually. In real life, it is a very time-consuming and impractical task for a physician to collect a large amount of patient data and provide the correct label. In order to solve the above problem, a liver image processing method is required in order to realize a classification mechanism that does not require the use of a large number of correct labels.
Disclosure of Invention
It is an object of the present invention to provide a liver image processing method which overcomes some or some of the disadvantages of the prior art.
The liver image processing method comprises the following steps:
firstly, extracting the characteristic of the image group from the CECT image of the liver part through A.K. software;
setting a threshold, and removing features lower than the threshold by using a VarianceThreshold method;
thirdly, performing feature dimension reduction by using a T-SNE dimension reduction algorithm;
fourthly, training and testing the training set samples by utilizing a QN-S3VM semi-supervised classification algorithm;
and fifthly, randomly distributing a test set and a training set for multiple times, and taking the average value of the classification accuracy of the test set and the training set.
Preferably, in the second step, a threshold is set first, then the variance of the sample feature is calculated, and if the variance of the feature is greater than the threshold, the feature is retained; if the variance of the feature is less than the threshold, the feature is deleted.
Preferably, the T-SNE dimension reduction algorithm maps the point pairs in the high-dimensional space to the low-dimensional space while keeping the distribution probability of the point pairs unchanged, converts the distance into the probability distribution by using Gaussian distribution in the high-dimensional space, converts the distance into the probability distribution by using T distribution in the low-dimensional space, represents the similarity of the point pairs by adopting joint probability, and obtains the sample distribution in the low-dimensional space by optimizing the KL divergence of the distance between the two distributions.
Preferably, the specific process of the T-SNE dimension reduction algorithm is as follows:
1) and setting the high-dimensional data point as X ═ X (X)1,x2,...,xn) The low-dimensional mapping point Y is (Y)1,y2,...,yn)；
2) The KL divergence is:
in the formula pijJoint probability function, q, for high dimensional spatial sample distributionijIs the joint probability of the sample distribution in the low dimensional space;
3) gaussian data point xi，xjThe joint probability function of (a) is as follows:
wherein σ is centered at xiThe optimal sigma is obtained by performing binary search on the Gaussian variance of the vector through a preset complexity factor;
4) in a low-dimensional space, data point yi，yjThe joint probability function of (a) is:
5) and finally, continuously updating and iterating by a gradient descent method to obtain a low-dimensional mapping sample.
Preferably, the flow of the QN-S3VM semi-supervised classification algorithm is as follows:
a. all labeled data, samples with hepatic failure labeled 1, samples without hepatic failure labeled-1, were scored as 2: 1, randomly distributing a training set and a testing set in proportion;
b. set the labeled sample of the training as Sl＝{(x1,y1′),...,(xm,ym') } unlabeled sample Su＝{(xm+1,ym+1′),...,(xn,yn') }, the number of marked samples is m, the number of unmarked samples is n-m, and the number of training lumped samples is n;
c. setting an initialization parameter lambda as a minimum iteration error, and setting S as a maximum iteration number;
d. recording the iteration times as s, iterating the training set sample in the step 1, calculating an error lambda' after each iteration, and adding 1 to the value of s after each iteration is finished;
e. until λ > ═ λ' or S = S, at which point the model training ends;
f. and placing the test set into a model for testing to obtain a prediction result.
The invention provides a liver image processing method based on a T-SNE characteristic reduction and QN-S3VM semi-supervised classification algorithm, which comprises the steps of firstly extracting CECT image texture characteristics of a liver part by utilizing A.K. software, then adding clinical physiological indexes, screening out characteristics with small variance through threshold value screening, mapping high-dimensional characteristics to a low-dimensional space by utilizing a dimension reduction algorithm T-SNE, reducing characteristic dimension, filtering out redundant characteristic information, finally randomly dividing a sample subjected to an operation into a training set and a test set through a semi-supervised learning idea, training the training set sample and a large number of patients subjected to preoperative examination, obtaining a final model, and predicting the accuracy of the test set sample.
The image omics is a new subject direction based on quantitative imaging, and converts medical image data into quantitative and high-dimensional data characteristics through high-flux extraction of quantitative characteristics. The liver function evaluation method is combined with clinical physiological indexes to be used for liver reserve function evaluation, multi-mode information complementation can be realized, deeper semantic information is mined, and the purpose of more accurately judging the condition of a testee is achieved.
In addition, the traditional supervised algorithm usually needs a large number of samples for training, and since medical data is precious, the acquisition of a large number of sample data is unrealistic.
Drawings
Fig. 1 is a flowchart of a liver image processing method according to embodiment 1;
FIG. 2 is a flowchart of feature selection in embodiment 1;
FIG. 3 is a flowchart of the QN-S3VM semi-supervised classification algorithm in example 1.
Detailed Description
For a further understanding of the invention, reference should be made to the following detailed description taken in conjunction with the accompanying drawings and examples. It is to be understood that the examples are illustrative of the invention and not limiting.
Example 1
As shown in fig. 1, the present embodiment provides a liver image processing method, which includes the following steps:
firstly, extracting the characteristic of the image group from the CECT image of the liver part through A.K. software;
setting a threshold, and removing features lower than the threshold by using a VarianceThreshold method;
thirdly, performing feature dimension reduction by using a T-SNE dimension reduction algorithm;
fourthly, training and testing the training set samples by utilizing a QN-S3VM semi-supervised classification algorithm;
and fifthly, randomly distributing a test set and a training set for multiple times, and taking the average value of the classification accuracy of the test set and the training set.
In fig. 1, the part enclosed by the left dotted line is mainly subjected to feature extraction and combination of the image omics features and the physiological features through a.k. software, and the part enclosed by the right dotted line is mainly: data preprocessing is carried out firstly, feature selection, data normalization and regularization are carried out according to a threshold value method, then data dimension reduction is carried out by using a T-SNE algorithm, and finally semi-supervised classification is carried out by using QN-S3 VM.
As shown in fig. 2, in the feature selection, a threshold is first set, then the variance of the sample feature is calculated, and if the variance of the feature is greater than the threshold, the feature is retained; if the variance of the feature is less than the threshold, the feature is deleted.
In this embodiment, the T-SNE dimension reduction algorithm maps the point pairs in the high-dimensional space to the low-dimensional space while keeping the distribution probability of each point pair unchanged, converts the distance into the probability distribution in the high-dimensional space using gaussian distribution, converts the distance into the probability distribution in the low-dimensional space using T distribution, represents the similarity of the point pairs using joint probability, and obtains the sample distribution in the low-dimensional space by optimizing the KL divergence of the distance between the two distributions.
In this embodiment, the T-SNE dimension reduction algorithm converts the distance into a probability distribution using Gaussian distribution in a high-dimensional space. Under a low-dimensional space, the similarity is expressed by using t distribution, so that points with low similarity are more distant, and points with high similarity are more compact, and the classification operation is favorably carried out later. The specific flow of the T-SNE dimension reduction algorithm is as follows:
1) and setting the high-dimensional data point as X ═ X (X)1,x2,...,xn) The low-dimensional mapping point Y is (Y)1,y2,...,yn)；
2) The KL divergence is:
in the formula pijJoint probability function, q, for high dimensional spatial sample distributionijIs the joint probability of the sample distribution in the low dimensional space;
3) gaussian data point xi，xjThe joint probability function of (a) is as follows:
wherein σ is centered at xiThe optimal sigma is obtained by performing binary search on the Gaussian variance of the vector through a preset complexity factor; k is in the interval [1, n ]]Taking any value;
4) in a low-dimensional space, data point yi，yjThe joint probability function of (a) is:
5) and finally, continuously updating and iterating by a gradient descent method to obtain a low-dimensional mapping sample.
In this embodiment, as shown in fig. 3, the flow of the QN-S3VM semi-supervised classification algorithm is as follows:
a. all labeled data, samples with hepatic failure labeled 1, samples without hepatic failure labeled-1, were scored as 2: 1, randomly distributing a training set and a testing set in proportion;
b. set the labeled sample of the training as Sl＝{(x1,y1′),...,(xm,ym') } unlabeled sample Su＝{(xm+1,ym+1′),...,(xn,yn') }, the number of marked samples is m, the number of unmarked samples is n-m, and the number of training lumped samples is n;
c. setting an initialization parameter lambda as a minimum iteration error, and setting S as a maximum iteration number;
d. recording the iteration times as s, iterating the training set sample in the step 1, calculating an error lambda' after each iteration, and adding 1 to the value of s after each iteration is finished;
e. until λ > ═ λ' or S = S, at which point the model training ends;
f. and placing the test set into a model for testing to obtain a prediction result.
Firstly, feature extraction is carried out on a liver part CECT image through A.K. software, then a threshold value is adjusted, features are screened through a VarianceThreshold method, and features lower than the threshold value are eliminated; converting the similarity between the data into probability by using a T-SNE dimension reduction algorithm, converting high-dimensional data into low-dimensional data by using a mapping relation, reducing the dimension of the data, and filtering out part of redundant information; the data to be tagged is represented by 2: 1, dividing the data into a training set and a test set, adding a large amount of label-free data into the training set, training by using a QN-S3VM algorithm, placing the test data into a trained model for testing to obtain a classification result, and performing multi-aspect evaluation on the model; and finally, randomly distributing a test set and a training set for many times, and taking the average value of the classification accuracy of the test set and the training set.
The T-SNE dimension reduction algorithm is a mature unsupervised dimension reduction algorithm, and maps a point pair of a high-dimensional space to a low-dimensional space according to the idea of converting distance into probability so as to reduce characteristic dimensions. When the semi-supervised model is trained, the initial minimum iteration error and the maximum iteration times are set. And when any one of the conditions is met in the iteration process, the model training is finished, and the model training can be used for predicting the test sample.
The present invention and its embodiments have been described above schematically, without limitation, and what is shown in the drawings is only one of the embodiments of the present invention, and the actual structure is not limited thereto. Therefore, if the person skilled in the art receives the teaching, without departing from the spirit of the invention, the person skilled in the art shall not inventively design the similar structural modes and embodiments to the technical solution, but shall fall within the scope of the invention.
Claims (5)
1. A liver image processing method characterized by: the method comprises the following steps:
firstly, extracting the characteristic of the image group from the CECT image of the liver part through A.K. software;
setting a threshold, and removing features lower than the threshold by using a VarianceThreshold method;
thirdly, performing feature dimension reduction by using a T-SNE dimension reduction algorithm;
fourthly, training and testing the training set samples by utilizing a QN-S3VM semi-supervised classification algorithm;
and fifthly, randomly distributing a test set and a training set for multiple times, and taking the average value of the classification accuracy of the test set and the training set.
2. A liver image processing method according to claim 1, characterized in that: in the second step, firstly setting a threshold, then calculating the variance of the sample characteristics, and if the variance of the characteristics is greater than the threshold, keeping the characteristics; if the variance of the feature is less than the threshold, the feature is deleted.
3. A liver image processing method according to claim 1, characterized in that: the T-SNE dimension reduction algorithm is characterized in that point pairs of a high-dimensional space are mapped to a low-dimensional space, the distribution probability of the point pairs is kept unchanged, the distance is converted into probability distribution in the high-dimensional space by using Gaussian distribution, the distance is converted into probability distribution in the low-dimensional space by using T distribution, the similarity of corresponding points is represented by adopting joint probability, and the sample distribution of the low-dimensional space is obtained by optimizing the KL divergence of the distance between the two distributions.
4. A liver image processing method according to claim 3, characterized in that: the specific flow of the T-SNE dimension reduction algorithm is as follows:
1) and setting the high-dimensional data point as X ═ X (X)1,x2,...,xn) The low-dimensional mapping point Y is (Y)1,y2,...,yn)；
2) The KL divergence is:
in the formula pijJoint probability function, q, for high dimensional spatial sample distributionijIs the joint probability of the sample distribution in the low dimensional space;
3) gaussian data point xi，xjThe joint probability function of (a) is as follows:
wherein σ is centered at xiThe optimal sigma is obtained by performing binary search on the Gaussian variance of the vector through a preset complexity factor;
4) in a low-dimensional space, data point yi，yjThe joint probability function of (a) is:
5) and finally, continuously updating and iterating by a gradient descent method to obtain a low-dimensional mapping sample.
5. A liver image processing method according to claim 1, characterized in that: the QN-S3VM semi-supervised classification algorithm comprises the following steps:
a. all labeled data, samples with hepatic failure labeled 1, samples without hepatic failure labeled-1, were scored as 2: 1, randomly distributing a training set and a testing set in proportion;
b. set the labeled sample of the training as Sl＝{(x1,y1′),...,(xm,ym') } unlabeled sample Su＝{(xm+1,ym+1′),...,(xn,yn') }, the number of marked samples is m, the number of unmarked samples is n-m, and the number of training lumped samples is n;
c. setting an initialization parameter lambda as a minimum iteration error, and setting S as a maximum iteration number;
d. recording the iteration times as s, iterating the training set sample in the step 1, calculating an error lambda' after each iteration, and adding 1 to the value of s after each iteration is finished;
e. until λ > ═ λ' or S = S, at which point the model training ends;
f. and placing the test set into a model for testing to obtain a prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110096604.0A CN112819765A (en) | 2021-01-25 | 2021-01-25 | Liver image processing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110096604.0A CN112819765A (en) | 2021-01-25 | 2021-01-25 | Liver image processing method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112819765A true CN112819765A (en) | 2021-05-18 |
Family
ID=75859054
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110096604.0A Pending CN112819765A (en) | 2021-01-25 | 2021-01-25 | Liver image processing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112819765A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115132352A (en) * | 2022-07-04 | 2022-09-30 | 西南交通大学 | Method for predicting curative effect of HIFU ablation hysteromyoma based on machine learning algorithm |
| CN117011242A (en) * | 2023-07-10 | 2023-11-07 | 珠海市人民医院 | Method and system for predicting hepatic encephalopathy after internal portal bypass operation through jugular vein |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130346346A1 (en) * | 2012-06-21 | 2013-12-26 | Microsoft Corporation | Semi-supervised random decision forests for machine learning |
| CN106530298A (en) * | 2016-11-14 | 2017-03-22 | 同济大学 | Three-way-decision-based liver tumor CT image classification method |
| CN108596204A (en) * | 2018-03-15 | 2018-09-28 | 西安电子科技大学 | A method of the semi-supervised modulation system disaggregated model based on modified SCDAE |
| CN109522973A (en) * | 2019-01-17 | 2019-03-26 | 云南大学 | Medical big data classification method and system based on production confrontation network and semi-supervised learning |
| CN110265095A (en) * | 2019-05-22 | 2019-09-20 | 首都医科大学附属北京佑安医院 | For HCC recurrence and construction method and the application of the prediction model and nomogram of RFS |
| US10430946B1 (en) * | 2019-03-14 | 2019-10-01 | Inception Institute of Artificial Intelligence, Ltd. | Medical image segmentation and severity grading using neural network architectures with semi-supervised learning techniques |
| CN110516688A (en) * | 2019-08-30 | 2019-11-29 | 北京推想科技有限公司 | The extracting method and system of Lung neoplasm attributive character information |
| CN110895817A (en) * | 2019-11-01 | 2020-03-20 | 复旦大学 | MRI image hepatic fibrosis automatic grading method based on image omics analysis |
| CN111127448A (en) * | 2019-12-26 | 2020-05-08 | 西南交通大学 | Model and method for detecting air spring fault based on isolated forest |
| CN111340685A (en) * | 2020-02-14 | 2020-06-26 | 中国地质大学（武汉） | Manifold dimension reduction method for remote sensing data processing |
| CN111488934A (en) * | 2020-04-13 | 2020-08-04 | 脑结构(南京)科技有限公司 | Brain image data processing method, storage medium, computer device and apparatus |
-
2021
- 2021-01-25 CN CN202110096604.0A patent/CN112819765A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130346346A1 (en) * | 2012-06-21 | 2013-12-26 | Microsoft Corporation | Semi-supervised random decision forests for machine learning |
| CN106530298A (en) * | 2016-11-14 | 2017-03-22 | 同济大学 | Three-way-decision-based liver tumor CT image classification method |
| CN108596204A (en) * | 2018-03-15 | 2018-09-28 | 西安电子科技大学 | A method of the semi-supervised modulation system disaggregated model based on modified SCDAE |
| CN109522973A (en) * | 2019-01-17 | 2019-03-26 | 云南大学 | Medical big data classification method and system based on production confrontation network and semi-supervised learning |
| US10430946B1 (en) * | 2019-03-14 | 2019-10-01 | Inception Institute of Artificial Intelligence, Ltd. | Medical image segmentation and severity grading using neural network architectures with semi-supervised learning techniques |
| CN110265095A (en) * | 2019-05-22 | 2019-09-20 | 首都医科大学附属北京佑安医院 | For HCC recurrence and construction method and the application of the prediction model and nomogram of RFS |
| CN110516688A (en) * | 2019-08-30 | 2019-11-29 | 北京推想科技有限公司 | The extracting method and system of Lung neoplasm attributive character information |
| CN110895817A (en) * | 2019-11-01 | 2020-03-20 | 复旦大学 | MRI image hepatic fibrosis automatic grading method based on image omics analysis |
| CN111127448A (en) * | 2019-12-26 | 2020-05-08 | 西南交通大学 | Model and method for detecting air spring fault based on isolated forest |
| CN111340685A (en) * | 2020-02-14 | 2020-06-26 | 中国地质大学（武汉） | Manifold dimension reduction method for remote sensing data processing |
| CN111488934A (en) * | 2020-04-13 | 2020-08-04 | 脑结构(南京)科技有限公司 | Brain image data processing method, storage medium, computer device and apparatus |
Non-Patent Citations (3)
| Title |
|---|
| FABIAN GIESEKE 等: "Fast and simple gradient-based optimization for semi-supervised support vector machines", 《NEUROCOMPUTING》 * |
| WEIHAO SUN 等: "QN-S3VM Method for Evaluation of Liver Functional Reserve", 《2020 CHINESE AUTOMATION CONGRESS (CAC)》 * |
| 郑洲 等: "基于区域增长与统一化水平集的CT肝脏图像分割", 《浙江大学学报(工学版)》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115132352A (en) * | 2022-07-04 | 2022-09-30 | 西南交通大学 | Method for predicting curative effect of HIFU ablation hysteromyoma based on machine learning algorithm |
| CN117011242A (en) * | 2023-07-10 | 2023-11-07 | 珠海市人民医院 | Method and system for predicting hepatic encephalopathy after internal portal bypass operation through jugular vein |
| CN117011242B (en) * | 2023-07-10 | 2024-05-14 | 珠海市人民医院 | Method and system for predicting hepatic encephalopathy after internal portal bypass operation through jugular vein |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jaworek-Korjakowska et al. | Melanoma thickness prediction based on convolutional neural network with VGG-19 model transfer learning | |
| CN108464840B (en) | Automatic detection method and system for breast lumps | |
| Mahapatra | Semi-supervised learning and graph cuts for consensus based medical image segmentation | |
| CN106056595B (en) | Based on the pernicious assistant diagnosis system of depth convolutional neural networks automatic identification Benign Thyroid Nodules | |
| CN110534192B (en) | Deep learning-based lung nodule benign and malignant recognition method | |
| CN111553892B (en) | Lung nodule segmentation calculation method, device and system based on deep learning | |
| Li et al. | Automatic lumbar spinal MRI image segmentation with a multi-scale attention network | |
| CN111598875A (en) | Method, system and device for building thyroid nodule automatic detection model | |
| WO2023207820A1 (en) | Pancreatic postoperative diabetes prediction system based on supervised deep subspace learning | |
| CN113077434B (en) | Method, device and storage medium for lung cancer identification based on multi-modal information | |
| CN115131642B (en) | Multi-modal medical data fusion system based on multi-view subspace clustering | |
| CN115497623A (en) | Lung cancer prognosis prediction system based on image, pathology and gene multiomics | |
| Ye et al. | Medical image diagnosis of prostate tumor based on PSP-Net+ VGG16 deep learning network | |
| JP2022547722A (en) | Weakly Supervised Multitask Learning for Cell Detection and Segmentation | |
| CN112819765A (en) | Liver image processing method | |
| CN111738997A (en) | Method for calculating new coronary pneumonia lesion area ratio based on deep learning | |
| CN109191422B (en) | System and method for detecting early ischemic stroke based on conventional CT image | |
| Jena et al. | Morphological feature extraction and KNG‐CNN classification of CT images for early lung cancer detection | |
| CN112508884A (en) | Comprehensive detection device and method for cancerous region | |
| CN113269799A (en) | Cervical cell segmentation method based on deep learning | |
| CN117036288A (en) | Tumor subtype diagnosis method for full-slice pathological image | |
| Wang et al. | Automatic measurement of fetal head circumference using a novel GCN-assisted deep convolutional network | |
| CN114519705A (en) | Ultrasonic standard data processing method and system for medical selection and identification | |
| CN116228759B (en) | Computer-aided diagnosis system and apparatus for renal cell carcinoma type | |
| CN117474876A (en) | Deep learning-based kidney cancer subtype auxiliary diagnosis and uncertainty evaluation method |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210518 |
|
| RJ01 | Rejection of invention patent application after publication |