CN112927203A - Glioma patient postoperative life prediction method based on multi-sequence MRI global information - Google Patents

Glioma patient postoperative life prediction method based on multi-sequence MRI global information Download PDF

Info

Publication number
CN112927203A
CN112927203A CN202110210990.1A CN202110210990A CN112927203A CN 112927203 A CN112927203 A CN 112927203A CN 202110210990 A CN202110210990 A CN 202110210990A CN 112927203 A CN112927203 A CN 112927203A
Authority
CN
China
Prior art keywords
prediction
age
months
data
sequence
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
Application number
CN202110210990.1A
Other languages
Chinese (zh)
Inventor
夏勇
蒋博文
张建鹏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Northwestern Polytechnical University
Shenzhen Institute of Northwestern Polytechnical University
Original Assignee
Northwestern Polytechnical University
Shenzhen Institute of Northwestern Polytechnical University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Northwestern Polytechnical University, Shenzhen Institute of Northwestern Polytechnical University filed Critical Northwestern Polytechnical University
Priority to CN202110210990.1A priority Critical patent/CN112927203A/en
Publication of CN112927203A publication Critical patent/CN112927203A/en
Pending legal-status Critical Current

Links

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
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/2431—Multiple classes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
    • 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
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G06T2207/10088—Magnetic resonance imaging [MRI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—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
    • G06T2207/20—Special algorithmic details
    • G06T2207/20112—Image segmentation details
    • G06T2207/20132—Image cropping
    • 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
    • G06T2207/30096—Tumor; Lesion

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Medical Informatics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Public Health (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Databases & Information Systems (AREA)
  • Pathology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)

Abstract

The invention provides a glioma patient postoperative life prediction method based on multi-sequence MRI global information, which is characterized in that after data preprocessing operation is carried out on images of four modes, splicing is carried out according to the sequence of T1, FLAIR, T2 and T1ce, prediction is carried out by adopting a 3D ResNet 50-based network model, and the global characteristics are weighted by introducing age characteristics in the prediction process. The glioma patient life-cycle three-dimensional regression prediction method based on the global information is convenient for learning the characteristics of tumor size, position, shape, texture, heterogeneity and the like after cutting the multi-sequence magnetic resonance image. A more ideal prediction result is obtained on a BraTS 2020 verification set.

Description

Glioma patient postoperative life prediction method based on multi-sequence MRI global information
Technical Field
The invention relates to the field of image analysis, in particular to a method for predicting the life cycle of a patient by using a nuclear magnetic resonance image.
Background
In recent years, the incidence of glioma is increasing year by year, and the glioma has high malignancy and high lethality, and is one of malignant tumors which seriously affect human health. Due to the difficulties of image degradation, complex pathological characterization of glioma imaging, obvious individual difference and the like, the survival prediction of glioma patients has great challenge. Most of the existing methods are based on the characteristics of manual design and adopt the traditional machine learning method. However, the extraction of manually designed features greatly depends on segmentation labels, and once the segmentation labels are inaccurate, the reliability of the features is inevitably reduced. Meanwhile, the rapid development of deep learning makes it widely used in the fields of computer vision and the like. The classic network model of deep learning: three-dimensional Residual Neural networks (3D resnets) are increasingly used in the field of medical image processing. Such as the Tencent medical team, uses a three-dimensional convolutional neural network-based medical Net network model to segment gliomas from multi-sequence magnetic resonance images.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides a glioma patient postoperative survival period prediction method based on multi-sequence MRI global information. In order to provide suggestions and references for doctor to make treatment schemes before operation and to accurately and effectively predict the prognosis survival time of glioma patients, the invention provides a glioma patient postoperative survival time prediction method based on multi-sequence magnetic resonance image global information fusion. In the training phase, the obtained data set includes age information of the patient and magnetic resonance images with four modalities of T1, T1ce, T2 and FLAIR. After data preprocessing operation is carried out on images of four modes, splicing is carried out according to the sequence of T1, FLAIR, T2 and T1ce, and prediction is carried out by adopting a network model based on 3D ResNet 50. Since age is related to patient prognostic recovery, an age feature is introduced in the prediction process to weight global features.
The technical scheme adopted by the invention for solving the technical problem comprises the following steps:
step 1, preprocessing data;
firstly, carrying out data normalization processing on voxels, and mapping data to a [0,1] interval, so as to facilitate network model learning, as shown in formula (1):
Figure BDA0002951442010000021
wherein x' represents the normalized voxel value, x represents the voxel value of the nuclear magnetic resonance image,
Figure BDA0002951442010000022
represents the mean of the non-zero voxel values, σ represents the standard deviation of the non-zero voxel values;
cutting off the black edge of the multi-sequence nuclear magnetic resonance image, scaling the image after cutting off the black edge to d h w size by adopting a nearest neighbor interpolation method, wherein d represents the depth, h represents the length, and w represents the width, and splicing according to the channels according to the sequence of T1, FLAIR, T2 and T1 ce;
step 2, regression prediction;
inputting a multi-sequence magnetic resonance image containing global information of tumor size, position, shape and texture as training data into a three-dimensional regression network model based on a 3D ResNet 50 for regression prediction;
updating model parameters by calculating a loss function and carrying out gradient back propagation; classifying results output by a prediction model, calculating a loss function for data with inconsistent categories when data with inconsistent prediction results and real situation categories exist, and performing gradient back propagation updating; calculating a loss function for all data in the batch when the prediction results of all data in the current batch are the same as the corresponding real situation categories, wherein the loss function is shown as a formula (5):
Figure BDA0002951442010000023
wherein the content of the first and second substances,
Figure BDA0002951442010000024
indicates the number of months to live, y, predicted when the categories are the sameiThe number of real life months when the types are the same,
Figure BDA0002951442010000025
the category is represented as the real survival month when the categories are inconsistent, y' represents the predicted number of survival months when the categories are inconsistent, class (·) represents the classification of the number of survival months, wherein the classification is respectively three categories of less than 10 months, more than or equal to 10 months, less than or equal to 15 months and more than 15 months, and the classification result is shown as formula (6):
Figure BDA0002951442010000026
step 3, data post-processing
According to the prediction result obtained by the prediction model, the number of the predicted survival months
Figure BDA0002951442010000027
Or
Figure BDA0002951442010000028
The number of days to live is multiplied by 30 to obtain the predicted number of days to live.
The three-dimensional regression network model knot based on the 3D ResNet 50 comprises 1 convolutional layer, 1 maximum pooling layer, 4 Bottleneck group modules, 1 global average pooling layer and 1 full-connection layer, wherein the convolutional kernel size of the convolutional layer is 7 × 7 × 7, the sliding step size is 2, the size of the obtained feature graph is reduced to be half of the original size, the sliding window size of the maximum pooling layer is 3 × 3 × 3, the step size is 2, the 4 Bottleneck group modules respectively comprise 3 Bottleneck modules, 4 Bottleneck modules, 6 Bottleneck modules and 3 Bottleneck modules, the Bottleneck modules comprise 1 jump connection layer, 3 convolutional layers, 3 batch one-way normalization layers and 3 ReLU activation functions, and the 3 volume layers comprise 2 volume layers with the step size of 1 × 1, the convolutional layer with the kernel size of 1 × 1 and 1 convolutional layer with the convolutional kernel size of 1 × 3 × 3.
Before the calculation of the full connection layer, introducing age characteristic factors, and after the global average pooling layer, multiplying a global characteristic graph obtained by a network model by the age characteristic element by element, wherein the structure of the age characteristic is shown as the following formula:
Fage＝1-age/100 (2)
wherein FageThe age characteristics are shown, and age is age information.
The activation function adopts a modified linear unit activation function (ReLU), as shown in formula (3):
f(x)＝max(0，x) (3)
wherein x represents an input element of the activation layer;
the Loss function adopts a mean square error Loss function (MSE Loss), and is shown in formula (4):
Figure BDA0002951442010000031
the invention has the beneficial effects that: many existing prediction methods are based on manual feature setting and manual extraction, and the method is time-consuming and labor-consuming. The glioma patient life-cycle three-dimensional regression prediction method based on the global information is convenient for learning the characteristics of tumor size, position, shape, texture, heterogeneity and the like after cutting the multi-sequence magnetic resonance image. A more ideal prediction result is obtained on a BraTS 2020 verification set.
Drawings
FIG. 1 is a three-dimensional regression network model structure diagram based on 3D ResNet 50.
Fig. 2 is a structural diagram of a bottleeck module according to the present invention.
Detailed Description
The invention is further illustrated with reference to the following figures and examples.
The invention relates to a glioma patient life cycle three-dimensional regression prediction method based on global information, which comprises the following specific steps:
step 1, preprocessing data;
because the voxel value range of the nuclear magnetic resonance image is large, firstly, the voxel is subjected to data normalization processing, and the data is mapped to a [0,1] interval, so that the network model learning is facilitated, as shown in formula (1):
Figure BDA0002951442010000041
wherein x' represents the normalized voxel value, x represents the voxel value of the nuclear magnetic resonance image,
Figure BDA0002951442010000042
represents the mean of the non-zero voxel values, σ represents the standard deviation of the non-zero voxel values;
the multi-sequence nuclear magnetic resonance image in the data set is larger than the outline of a brain, a black edge exists around the multi-sequence nuclear magnetic resonance image, so that the waste of video memory is caused, the black edge needs to be cut, the image after the black edge is cut is zoomed to d x h w size by adopting a nearest neighbor interpolation method, wherein d represents the depth, h represents the length, w represents the width, and the multi-sequence nuclear magnetic resonance image is spliced according to the channels according to the sequence of T1, FLAIR, T2 and T1 ce;
step 2, regression prediction;
inputting a multi-sequence magnetic resonance image containing global information of tumor size, position, shape and texture as training data into a three-dimensional regression network model based on a 3D ResNet 50 for regression prediction; the structure of the three-dimensional regression network model based on the 3D ResNet 50 is shown in the following FIG. 1;
the three-dimensional regression network model structure comprises 1 convolutional layer, 1 maximum pooling layer, 4 Bottleneck group modules, 1 global average pooling layer and 1 full-connection layer, wherein the convolutional kernel size of the convolutional layer is 7 × 7 × 7, the sliding step length is 2, the size of an obtained feature graph is reduced to be half of the original size, the sliding window size of the maximum pooling layer is 3 × 3 × 3, the step length is 2, the 4 Bottleneck group modules respectively comprise 3 Bottleneck modules, 4 Bottleneck modules, 6 Bottleneck modules and 3 Bottleneck modules, the Bottleneck modules comprise 1 jump connection, 3 convolutional layers, 3 batch convolutional layer normalization layers and 3 Reactivating functions, and the 3 convolutional layers comprise 2 convolutional layers with the step length of 1, the convolutional kernel size of 1 × 1 × 1 and 1 convolutional layer with the step length of 1 × 3 × 3 convolutional kernel size;
before the calculation of the full connection layer, considering that the age is related to the prognosis recovery condition of the patient, an age characteristic factor is introduced, and after the global average pooling layer, a global characteristic graph obtained by a network model is multiplied by the age characteristic element by element, wherein the structure of the age characteristic is shown as the following formula:
Fage＝1-age/100 (2)
wherein FageRepresenting age characteristics, age representing age information;
the activation function is a modified linear unit activation function (ReLU), as shown in equation (3):
f(x)＝max(0，x) (3)
wherein x represents an input element of the activation layer;
the Loss function adopts a mean square error Loss function (MSE Loss), and is shown in formula (4):
Figure BDA0002951442010000051
calculating a loss function by using a formula (5), and updating model parameters by gradient back propagation; classifying results output by a prediction model, calculating a loss function for data with inconsistent categories when data with inconsistent prediction results and real situation categories exist, and performing gradient back propagation updating; when the prediction results of all data in the current batch are the same as the corresponding real situation categories, calculating loss functions for all data in the batch, wherein the action effect is shown as the formula (5):
Figure BDA0002951442010000052
wherein the content of the first and second substances,
Figure BDA0002951442010000053
indicates the number of months of survival predicted, yiThe class (·) represents the number of actual survival months, and the class (·) represents the classification of the number of survival months, and is divided into three categories of less than 10 months, not less than 15 months, and more than 15 months, and the action and effect are shown as the following formula 6.
Figure BDA0002951442010000054
Step 3, data post-processing
And (4) multiplying the predicted survival month number by 30 according to the prediction result obtained by the prediction model, and converting into the predicted survival days.

Claims (5)

1. A glioma patient postoperative survival prediction method based on multi-sequence MRI global information is characterized by comprising the following steps:
step 1, preprocessing data;
firstly, carrying out data normalization processing on voxels, and mapping data to a [0,1] interval, so as to facilitate network model learning, as shown in formula (1):
Figure FDA0002951442000000011
wherein x' represents the normalized voxel value, x represents the voxel value of the nuclear magnetic resonance image,
Figure FDA0002951442000000012
represents the mean of the non-zero voxel values, σ represents the standard deviation of the non-zero voxel values;
cutting off the black edge of the multi-sequence nuclear magnetic resonance image, scaling the image after cutting off the black edge to d h w size by adopting a nearest neighbor interpolation method, wherein d represents the depth, h represents the length, and w represents the width, and splicing according to the channels according to the sequence of T1, FLAIR, T2 and T1 ce;
step 2, regression prediction;
inputting a multi-sequence magnetic resonance image containing global information of tumor size, position, shape and texture as training data into a three-dimensional regression network model based on a 3D ResNet 50 for regression prediction;
updating model parameters by calculating a loss function and carrying out gradient back propagation; classifying results output by a prediction model, calculating a loss function for data with inconsistent categories when data with inconsistent prediction results and real situation categories exist, and performing gradient back propagation updating; calculating a loss function for all data in the batch when the prediction results of all data in the current batch are the same as the corresponding real situation categories, wherein the loss function is shown as a formula (5):
Figure FDA0002951442000000013
wherein the content of the first and second substances,
Figure FDA0002951442000000014
indicates the number of months to live, y, predicted when the categories are the sameiThe number of real life months when the types are the same,
Figure FDA0002951442000000015
the category is represented as the real survival month when the categories are inconsistent, y' represents the predicted number of survival months when the categories are inconsistent, class (·) represents the classification of the number of survival months, wherein the classification is respectively three categories of less than 10 months, more than or equal to 10 months, less than or equal to 15 months and more than 15 months, and the classification result is shown as formula (6):
Figure FDA0002951442000000016
step 3, data post-processing
According to the prediction result obtained by the prediction model, the number of the predicted survival months
Figure FDA0002951442000000021
Or
Figure FDA0002951442000000022
The number of days to live is multiplied by 30 to obtain the predicted number of days to live.
2. The method for predicting the postoperative survival of a glioma patient based on multi-sequence MRI global information according to claim 1, wherein the method comprises the following steps:
the three-dimensional regression network model knot based on the 3D ResNet 50 comprises 1 convolutional layer, 1 maximum pooling layer, 4 Bottleneck group modules, 1 global average pooling layer and 1 full-connection layer, wherein the convolutional kernel size of the convolutional layer is 7 × 7 × 7, the sliding step size is 2, the size of the obtained feature graph is reduced to be half of the original size, the sliding window size of the maximum pooling layer is 3 × 3 × 3, the step size is 2, the 4 Bottleneck group modules respectively comprise 3 Bottleneck modules, 4 Bottleneck modules, 6 Bottleneck modules and 3 Bottleneck modules, the Bottleneck modules comprise 1 jump connection layer, 3 convolutional layers, 3 batch one-way normalization layers and 3 ReLU activation functions, and the 3 volume layers comprise 2 volume layers with the step size of 1 × 1, the convolutional layer with the kernel size of 1 × 1 and 1 convolutional layer with the convolutional kernel size of 1 × 3 × 3.
3. The method for predicting the postoperative survival of a glioma patient based on multi-sequence MRI global information according to claim 1, wherein the method comprises the following steps:
before the calculation of the full connection layer, introducing age characteristic factors, and after the global average pooling layer, multiplying a global characteristic graph obtained by a network model by the age characteristic element by element, wherein the structure of the age characteristic is shown as the following formula:
Fage＝1-age/100 (2)
wherein FageIndicating the age characteristic, age represents age information.
4. The method for predicting the postoperative survival of a glioma patient based on multi-sequence MRI global information according to claim 1, wherein the method comprises the following steps:
the activation function adopts a modified linear unit activation function (ReLU), as shown in formula (3):
f(x)＝max(0，x) (3)
where x represents an input element of the active layer.
5. The method for predicting the postoperative survival of a glioma patient based on multi-sequence MRI global information according to claim 1, wherein the method comprises the following steps:
the loss function adopts a mean square error loss function, and is shown in formula (4):
Figure FDA0002951442000000023
wherein n is the number of data.
CN202110210990.1A 2021-02-25 2021-02-25 Glioma patient postoperative life prediction method based on multi-sequence MRI global information Pending CN112927203A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110210990.1A CN112927203A (en) 2021-02-25 2021-02-25 Glioma patient postoperative life prediction method based on multi-sequence MRI global information

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110210990.1A CN112927203A (en) 2021-02-25 2021-02-25 Glioma patient postoperative life prediction method based on multi-sequence MRI global information

Publications (1)

Publication Number Publication Date
CN112927203A true CN112927203A (en) 2021-06-08

Family

ID=76171864

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110210990.1A Pending CN112927203A (en) 2021-02-25 2021-02-25 Glioma patient postoperative life prediction method based on multi-sequence MRI global information

Country Status (1)

Country Link
CN (1) CN112927203A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113689382A (en) * 2021-07-26 2021-11-23 北京知见生命科技有限公司 Tumor postoperative life prediction method and system based on medical images and pathological images

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108492297A (en) * 2017-12-25 2018-09-04 重庆理工大学 The MRI brain tumors positioning for cascading convolutional network based on depth and dividing method in tumor
CN108766563A (en) * 2018-05-25 2018-11-06 戴建荣 Radiotherapy prediction of result method and system based on dosage group
CN112184658A (en) * 2020-09-24 2021-01-05 上海健康医学院 Method, medium, and electronic device for predicting non-small cell lung cancer prognostic survival

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108492297A (en) * 2017-12-25 2018-09-04 重庆理工大学 The MRI brain tumors positioning for cascading convolutional network based on depth and dividing method in tumor
CN108766563A (en) * 2018-05-25 2018-11-06 戴建荣 Radiotherapy prediction of result method and system based on dosage group
CN112184658A (en) * 2020-09-24 2021-01-05 上海健康医学院 Method, medium, and electronic device for predicting non-small cell lung cancer prognostic survival

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
陈威: "基于深度学习与影像组学的癌症影像分析", 《中国优秀博士论文全文数据库医药卫生科技辑》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113689382A (en) * 2021-07-26 2021-11-23 北京知见生命科技有限公司 Tumor postoperative life prediction method and system based on medical images and pathological images
CN113689382B (en) * 2021-07-26 2023-12-01 北京知见生命科技有限公司 Tumor postoperative survival prediction method and system based on medical images and pathological images

Similar Documents

Publication Publication Date Title
Bi et al. Automatic liver lesion detection using cascaded deep residual networks
US20210012486A1 (en) Image synthesis with generative adversarial network
CN110889852B (en) Liver segmentation method based on residual error-attention deep neural network
CN111612754B (en) MRI tumor optimization segmentation method and system based on multi-modal image fusion
US10853409B2 (en) Systems and methods for image search
CN110992351B (en) sMRI image classification method and device based on multi-input convolution neural network
CN110706214B (en) Three-dimensional U-Net brain tumor segmentation method fusing condition randomness and residual error
CN104484886B (en) A kind of dividing method and device of MR images
Kim et al. Bidirectional rnn-based few shot learning for 3d medical image segmentation
CN111598871B (en) Multi-feature fusion assisted pulmonary glass shadow nodule detection system and medium
CN114782350A (en) Multi-modal feature fusion MRI brain tumor image segmentation method based on attention mechanism
CN109118487A (en) Bone age assessment method based on non-down sampling contourlet transform and convolutional neural networks
CN110930378A (en) Emphysema image processing method and system based on low data demand
CN116071383A (en) Hippocampus subzone segmentation method and system based on ultra-high field magnetic resonance image reconstruction
CN115496720A (en) Gastrointestinal cancer pathological image segmentation method based on ViT mechanism model and related equipment
CN114119516A (en) Virus focus segmentation method based on transfer learning and cascade adaptive hole convolution
CN112420170A (en) Method for improving image classification accuracy of computer aided diagnosis system
Agarwal et al. Weakly supervised lesion co-segmentation on ct scans
CN116664590B (en) Automatic segmentation method and device based on dynamic contrast enhancement magnetic resonance image
CN117437423A (en) Weak supervision medical image segmentation method and device based on SAM collaborative learning and cross-layer feature aggregation enhancement
CN112927203A (en) Glioma patient postoperative life prediction method based on multi-sequence MRI global information
Ullah et al. DSFMA: Deeply supervised fully convolutional neural networks based on multi-level aggregation for saliency detection
CN106709921B (en) Color image segmentation method based on space Dirichlet mixed model
Yan et al. Two and multiple categorization of breast pathological images by transfer learning
CN113409324B (en) Brain segmentation method fusing differential geometric information

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
RJ01 Rejection of invention patent application after publication

Application publication date: 20210608