CN113780274A - Method, system and medium for predicting liver cancer recurrence by combining imaging omics and deep learning - Google Patents
Method, system and medium for predicting liver cancer recurrence by combining imaging omics and deep learning Download PDFInfo
- Publication number
- CN113780274A CN113780274A CN202110982206.9A CN202110982206A CN113780274A CN 113780274 A CN113780274 A CN 113780274A CN 202110982206 A CN202110982206 A CN 202110982206A CN 113780274 A CN113780274 A CN 113780274A
- Authority
- CN
- China
- Prior art keywords
- liver cancer
- dimensional
- model
- prediction
- image
- 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
- 201000007270 liver cancer Diseases 0.000 title claims abstract description 74
- 208000014018 liver neoplasm Diseases 0.000 title claims abstract description 74
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000003384 imaging method Methods 0.000 title claims abstract description 30
- 238000013135 deep learning Methods 0.000 title claims abstract description 28
- 238000013145 classification model Methods 0.000 claims abstract description 20
- 238000003062 neural network model Methods 0.000 claims abstract description 17
- 238000013527 convolutional neural network Methods 0.000 claims description 15
- 238000011176 pooling Methods 0.000 claims description 15
- 238000012545 processing Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 5
- 230000009467 reduction Effects 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 3
- 230000009466 transformation Effects 0.000 claims description 3
- 230000010365 information processing Effects 0.000 abstract description 2
- 238000002591 computed tomography Methods 0.000 description 14
- 206010073071 hepatocellular carcinoma Diseases 0.000 description 7
- 238000004590 computer program Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 230000011218 segmentation Effects 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000003187 abdominal effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000012512 characterization method Methods 0.000 description 2
- 210000002767 hepatic artery Anatomy 0.000 description 2
- 238000001356 surgical procedure Methods 0.000 description 2
- 238000002560 therapeutic procedure Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 238000012752 Hepatectomy Methods 0.000 description 1
- 206010028980 Neoplasm Diseases 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 201000011510 cancer Diseases 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 231100000844 hepatocellular carcinoma Toxicity 0.000 description 1
- 210000004185 liver Anatomy 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000010837 poor prognosis Methods 0.000 description 1
- 238000004393 prognosis Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000002054 transplantation Methods 0.000 description 1
Images
Classifications
-
- 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
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a method, a system and a storage medium for predicting liver cancer recurrence by combining imaging omics and deep learning, wherein the method comprises the following steps: acquiring a case image of a marked liver cancer target region; inputting the case image into a prediction model to obtain a prediction result; the prediction model is obtained by the following steps: acquiring a case image sample of a marked liver cancer target region, and extracting an interested region image sample from the sample; extracting the image omics characteristics and the three-dimensional small blocks of the sample; inputting the image omics characteristics into a classification model to obtain omics scores, and inputting the three-dimensional small blocks into a three-dimensional convolution neural network model to obtain network scores; and inputting the omic score and the network score into the combined model to obtain a prediction score, determining a prediction result according to the prediction score, and determining the prediction model according to the accuracy of the prediction result. The embodiment of the invention can improve the accuracy of predicting the recurrence of the liver cancer, and can be widely applied to the technical field of medical data information processing.
Description
Technical Field
The invention relates to the technical field of medical data information processing, in particular to a method, a system and a storage medium for predicting liver cancer recurrence through combination of imaging and omics and deep learning.
Background
Primary Hepatocellular carcinoma (HCC, hereinafter referred to as liver cancer) is one of the most common malignant tumors worldwide. Currently, the most promising cure for liver cancer is liver transplantation and radical hepatectomy. However, the recurrence rate of surgical treatment for 5 years is as high as 18-72%, the recurrence after radical hepatoma therapy is a main factor of poor prognosis of patients with hepatoma, and how to effectively predict the recurrence after the hepatoma therapy is a difficult point. Therefore, it is very necessary to screen the high-risk recurrence population after the radical hepatoma treatment. How to deeply mine the imaging information and provide prognosis and curative effect information for patients is very important. The prior art is limited to artificial imagery omics characteristics extracted based on CT images, and is easily influenced by machine parameters, so that the prediction accuracy is reduced.
Disclosure of Invention
In view of this, an object of the embodiments of the present invention is to provide a method, a system and a storage medium for predicting liver cancer recurrence through combining imaging and deep learning, which can improve the accuracy of predicting liver cancer recurrence.
In a first aspect, an embodiment of the present invention provides a method for predicting liver cancer recurrence through combination of imaging and deep learning, including the following steps:
acquiring a case image of a marked liver cancer target region;
inputting the case image of the marked liver cancer target area into a prediction model to obtain a prediction result;
the prediction model comprises a classification model, a three-dimensional convolution neural network model and a joint model, and is obtained through the following steps:
acquiring a case image sample of a marked liver cancer target region, and extracting an interested region image sample from the case image sample of the marked liver cancer target region;
extracting the image omics characteristics and three-dimensional small blocks of the image sample of the region of interest;
inputting the image omics characteristics into the classification model to obtain omics scores, and inputting the three-dimensional small blocks into the three-dimensional convolution neural network model to obtain network scores;
and inputting the omic score and the network score into the combined model to obtain a prediction score, determining a prediction result according to the prediction score, and determining the prediction model according to the accuracy of the prediction result.
Optionally, before extracting the omics features of the image samples of the region of interest, the method further includes:
and carrying out first standardization processing and three-dimensional reconstruction on the image sample of the region of interest.
Optionally, before inputting the imagery omics features into the classification model to obtain an omics score, the method further comprises:
and performing dimension reduction processing on the image omics characteristics.
Optionally, the three-dimensional small block is obtained by:
performing a second normalization process on the region-of-interest image sample;
extracting a plurality of three-dimensional small blocks for each image sample of the region of interest.
Optionally, before inputting the three-dimensional small block into the three-dimensional convolutional neural network model to obtain a network score, the method further includes:
and performing data enhancement on the three-dimensional small blocks.
Optionally, the performing data enhancement on the three-dimensional small block specifically includes:
and turning, rotating, clipping, deforming, scaling, denoising, blurring, color transformation, erasing or filling the three-dimensional small blocks.
Optionally, the three-dimensional convolutional neural network model includes a first convolutional layer, a second convolutional layer, a first pooling layer, a third convolutional layer, a fourth convolutional layer, a first pooling layer, a first fully-connected layer, a second fully-connected layer, a third fully-connected layer, a fourth fully-connected layer, and a classifier, which are connected in sequence, and the inputting the three-dimensional small block into the three-dimensional convolutional neural network model to obtain a network score specifically includes:
and sequentially passing the three-dimensional small blocks through a first convolution layer, a second convolution layer, a first pooling layer, a third convolution layer, a fourth convolution layer, a first pooling layer, a first full-connection layer, a second full-connection layer, a third full-connection layer, a fourth full-connection layer and a classifier, and outputting a network score.
In a second aspect, an embodiment of the present invention provides a system for predicting liver cancer recurrence through imaging and omics combined deep learning, including:
the acquisition module is used for acquiring a case image of the marked liver cancer target area;
the prediction module is used for inputting the case image of the marked liver cancer target area into a prediction model to obtain a prediction result;
the prediction model comprises a classification model, a three-dimensional convolution neural network model and a joint model, and is obtained through the following steps:
acquiring a case image sample of a marked liver cancer target region, and extracting an interested region image sample from the case image sample of the marked liver cancer target region;
extracting the image omics characteristics and three-dimensional small blocks of the image sample of the region of interest;
inputting the image omics characteristics into the classification model to obtain omics scores, and inputting the three-dimensional small blocks into the three-dimensional convolution neural network model to obtain network scores;
and inputting the omic score and the network score into the combined model to obtain a prediction score, determining a prediction result according to the prediction score, and determining the prediction model according to the accuracy of the prediction result.
In a third aspect, an embodiment of the present invention provides a system for predicting liver cancer recurrence through imaging omics combined deep learning, including:
at least one processor;
at least one memory for storing at least one program;
when executed by the at least one processor, cause the at least one processor to implement the method described above.
In a fourth aspect, embodiments of the present invention provide a storage medium in which a processor-executable program is stored, the processor-executable program being configured to perform the above method when executed by a processor.
The implementation of the embodiment of the invention has the following beneficial effects: firstly training a prediction model according to case image sample data of a marked liver cancer target region, then inputting the case image of the marked liver cancer target region into the trained prediction model to obtain a prediction result, wherein the prediction model comprises a classification model, a three-dimensional convolution neural network model and a joint model, obtaining omic scores of image omic characteristics according to the classification model, obtaining deep learning network scores according to the three-dimensional convolution neural network model, inputting the omic scores and the network scores into the joint model to obtain prediction scores, and obtaining the prediction result according to the prediction scores; the prediction model combines the imaging omics and the deep learning, the mined image information is richer, and the characterization capability is stronger, so that the accuracy of predicting the recurrence of the liver cancer is improved.
Drawings
Fig. 1 is a schematic flowchart illustrating steps of a method for predicting liver cancer recurrence through imaging group and deep learning in combination according to an embodiment of the present invention;
FIG. 2 is a schematic structural diagram of a three-dimensional convolutional neural network model according to an embodiment of the present invention;
FIG. 3 is a chart of a case of a patient with liver cancer according to an embodiment of the present invention;
fig. 4 is a block diagram of a system for predicting recurrence of liver cancer by combining imaging and deep learning according to an embodiment of the present invention;
fig. 5 is another block diagram of a system for predicting recurrence of liver cancer by using imaging group in combination with deep learning according to an embodiment of the present invention.
Detailed Description
The invention is described in further detail below with reference to the figures and the specific embodiments. The step numbers in the following embodiments are provided only for convenience of illustration, the order between the steps is not limited at all, and the execution order of each step in the embodiments can be adapted according to the understanding of those skilled in the art.
As shown in fig. 1, the embodiment of the present invention provides a method for predicting recurrence of liver cancer by combining imaging group and deep learning, which comprises the following steps.
S100, acquiring a case image of the marked liver cancer target area;
s200, inputting the case image of the marked liver cancer target area into a prediction model to obtain a prediction result;
the prediction model comprises a classification model, a three-dimensional convolution neural network model and a joint model, and is obtained through the following steps:
acquiring a case image sample of a marked liver cancer target region, and extracting an interested region image sample from the case image sample of the marked liver cancer target region;
extracting the image omics characteristics and three-dimensional small blocks of the image sample of the region of interest;
inputting the image omics characteristics into the classification model to obtain omics scores, and inputting the three-dimensional small blocks into the three-dimensional convolution neural network model to obtain network scores;
and inputting the omic score and the network score into the combined model to obtain a prediction score, determining a prediction result according to the prediction score, and determining the prediction model according to the accuracy of the prediction result.
Specifically, a prediction model is trained according to sample data of a case image with a labeled liver cancer target area, and then the case image with the labeled liver cancer target area is input into the trained prediction model to obtain a prediction result.
The image of the labeled target area of liver cancer is obtained by the following method: selecting a liver cancer case meeting the requirement, downloading a Computed Tomography (CT) image of a hepatic artery stage through a PACS system, delineating a liver cancer target region by an abdominal radiologist through an ITK-SNAP, and respectively storing the liver cancer target region as a Main image (Main image) and a Segmentation image (Segmentation image) in an nii format.
Specifically, Logistic linear analysis is adopted for the imaging omics and the deep learning model, a joint prediction model is established, the joint score is a + b × omics score + c × network score, the score below 0 is low-risk recurrence after radical hepatoma surgery, and the score above 0 is high-risk recurrence.
Optionally, before extracting the omics features of the image samples of the region of interest, the method further includes:
s201, carrying out first standardization processing and three-dimensional reconstruction on the image sample of the region of interest.
Specifically, firstly, a first normalization process and a three-dimensional reconstruction are performed on a segmented three-dimensional ROI (Region Of Interest) image by using MATLAB2014b, and the thickness Of the reconstructed image layer is 1 mm; then, an image omics feature extraction package is installed based on Python3.6, and three-dimensional image omics features are extracted in batches.
Optionally, before inputting the imagery omics features into the classification model to obtain an omics score, the method further comprises:
s202, performing dimension reduction processing on the image omics characteristics.
Specifically, Recursive Feature Elimination (RFE) is used for dimensionality reduction, and a model is built by random construction. Defining an omics scoring (Rad-score) formula according to the weight of each image omics characteristic of the model, and then constructing an image omics label (Radiomics signature) through the Rad-score. And aiming at predicting the recurrence of the liver cancer, determining the optimal intercept value of the Rad-score by adopting the maximum johnson index.
Optionally, the three-dimensional small block is obtained by:
s203, carrying out second standardization processing on the image sample of the region of interest;
and S204, extracting a plurality of three-dimensional small blocks for each image sample of the region of interest.
Specifically, the process of extracting three-dimensional Patches (3D-Patches) is as follows: cropping the pre-processed volume number (Volumetric data) into 3D voxels 50 × 50 × 50 (voxel, 1 voxel represents 1 mm); the pretreatment follows the procedure of "standard" abdominal CT: the input CT image scan is converted to Hounsfield Units (HU), the volume data is then adjusted to a 1mm x 1mm spacing by linear interpolation, the voxel intensities are compressed to ihu e-1024, 400, the densities are quantized to gray scale, and the values are converted to I e-1, 1 by mapping. On the basis of CT image standardization, 10 three-dimensional small blocks (Patches) are extracted from each tumor three-dimensional ROI area
Optionally, before inputting the three-dimensional small block into the three-dimensional convolutional neural network model to obtain a network score, the method further includes:
and S205, performing data enhancement on the three-dimensional small blocks.
Optionally, the performing data enhancement on the three-dimensional small block specifically includes:
s206, turning, rotating, cutting, deforming, scaling, denoising, blurring, color transformation, erasing or filling the three-dimensional small blocks.
Specifically, the training data is augmented in 10 enhancements including, but not limited to, those described above.
Optionally, the three-dimensional convolutional neural network model includes a first convolutional layer, a second convolutional layer, a first pooling layer, a third convolutional layer, a fourth convolutional layer, a first pooling layer, a first fully-connected layer, a second fully-connected layer, a third fully-connected layer, a fourth fully-connected layer, and a classifier, which are connected in sequence, and the inputting the three-dimensional small block into the three-dimensional convolutional neural network model to obtain a network score specifically includes:
and sequentially passing the three-dimensional small blocks through a first convolution layer, a second convolution layer, a first pooling layer, a third convolution layer, a fourth convolution layer, a first pooling layer, a first full-connection layer, a second full-connection layer, a third full-connection layer, a fourth full-connection layer and a classifier, and outputting a network score.
Specifically, as shown in fig. 2, the architecture of a Three-dimensional convolutional neural network (3D-CNN) is as follows: four 3D Convolutional layers (connected layers), two pooling layers, four full-link layers and one classifier; filters (Filters) of the four 3D convolutional layers are 64, 128, 256 and 512, respectively, and Kernel sizes (Kernel sizes) thereof are 5 × 5 × 5, 3 × 3 × 3 and 3 × 3 × 3, respectively; two largest pooling layers with kernel sizes of 3 × 3 × 3 are arranged between the second convolution layer and the third convolution layer and behind the fourth convolution layer; four Fully connected layers (Fully connected layers) are arranged behind the second largest pooling layer, and the unit numbers of the four Fully connected layers are 13824, 512, 256 and 2 respectively; finally, a Softmax classifier is connected. The Optimizer of the three-dimensional convolutional neural network (Optimizer) selects Adam with a Learning rate (Learning rate) set to 0.001 and a Batch size (Batch size) of 16. All layers were normalized, with an L2 regularization penalty set to 0.000001, using a relu (rectified Linear units) activation function with an alpha of 0.1, trained and parametrized based on the tensrflow.
It should be noted that, the three-dimensional convolutional neural network has better stability as proved by experiments.
The implementation of the embodiment of the invention has the following beneficial effects: firstly training a prediction model according to case image sample data of a marked liver cancer target region, then inputting the case image of the marked liver cancer target region into the trained prediction model to obtain a prediction result, wherein the prediction model comprises a classification model, a three-dimensional convolution neural network model and a joint model, obtaining omic scores of image omic characteristics according to the classification model, obtaining deep learning network scores according to the three-dimensional convolution neural network model, inputting the omic scores and the network scores into the joint model to obtain prediction scores, and obtaining the prediction result according to the prediction scores; the prediction model combines the imaging omics and the deep learning, the mined image information is richer, and the characterization capability is stronger, so that the accuracy of predicting the recurrence of the liver cancer is improved.
The method for predicting recurrence of liver cancer by using imaging group and deep learning is described in a specific embodiment.
As shown in fig. 3, first, the three-dimensional CT image reconstruction process is as follows: firstly, delineating a liver cancer target region of a CT image of a liver cancer patient before operation layer by layer through ITK-SNAP, and respectively storing the CT image as a Main image (Main image) and a Segmentation image (Segmentation image) in a nii format as shown in T3-1; performing standardization and three-dimensional reconstruction on the segmented three-dimensional ROI image by using MATLAB2014 b; t3-2 is a 3D image generated after the target area is sketched for the liver cancer, and is a CT image with the thickness of 5 mm; t3-3 is a CT image reconstructed by matlab2014b to be 1mm layer thickness at the target region of 3D liver cancer.
Then, extracting three-dimensional CT image omics characteristics and calculating a score: for the hepatic artery phase CT image before treatment, standardized treatment and three-dimensional reconstruction (layer thickness is 1mm) are carried out, 1153 image omics characteristics are extracted by utilizing Python3.6, 14 image omics characteristics are obtained by RFE dimension reduction treatment, 14 image omics characteristic numerical values are calculated, and as shown in the following table, the omics score is 0.860 through trained random forest calculation.
Then, the CT image is input to the trained 3D-CNN to obtain a network score of 0.492.
And finally, inputting the omics score of 0.860 and the network score of 0.492 into the joint model to obtain a prediction score, wherein the joint model is determined as follows after training: the prediction score is 1.979731+2.425129 x omics score +1.987351 x network score; therefore, the prediction score is 5.037, the prediction score is more than 0, and the prediction result of the case is that the curative effect of the liver cancer radical treatment is poor and the liver cancer is easy to relapse.
As shown in fig. 4, an embodiment of the present invention provides a system for predicting liver cancer recurrence through imaging group combined deep learning, including:
the acquisition module is used for acquiring a case image of the marked liver cancer target area;
the prediction module is used for inputting the case image of the marked liver cancer target area into a prediction model to obtain a prediction result;
the prediction model comprises a classification model, a three-dimensional convolution neural network model and a joint model, and is obtained through the following steps:
acquiring a case image sample of a marked liver cancer target region, and extracting an interested region image sample from the case image sample of the marked liver cancer target region;
extracting the image omics characteristics and three-dimensional small blocks of the image sample of the region of interest;
inputting the image omics characteristics into the classification model to obtain omics scores, and inputting the three-dimensional small blocks into the three-dimensional convolution neural network model to obtain network scores;
and inputting the omic score and the network score into the combined model to obtain a prediction score, determining a prediction result according to the prediction score, and determining the prediction model according to the accuracy of the prediction result.
It can be seen that the contents in the foregoing method embodiments are all applicable to this system embodiment, the functions specifically implemented by this system embodiment are the same as those in the foregoing method embodiment, and the advantageous effects achieved by this system embodiment are also the same as those achieved by the foregoing method embodiment.
As shown in fig. 5, an embodiment of the present invention provides a system for predicting liver cancer recurrence through imaging group combined deep learning, including:
at least one processor;
at least one memory for storing at least one program;
when executed by the at least one processor, cause the at least one processor to implement the method described above.
It can be seen that the contents in the foregoing method embodiments are all applicable to this system embodiment, the functions specifically implemented by this system embodiment are the same as those in the foregoing method embodiment, and the advantageous effects achieved by this system embodiment are also the same as those achieved by the foregoing method embodiment.
In addition, the embodiment of the application also discloses a computer program product or a computer program, and the computer program product or the computer program is stored in a computer readable storage medium. The computer program may be read by a processor of a computer device from a computer-readable storage medium, and the computer program is executed by the processor to cause the computer device to perform the above-described method. Likewise, the contents of the above method embodiments are all applicable to the present storage medium embodiment, the functions specifically implemented by the present storage medium embodiment are the same as those of the above method embodiments, and the advantageous effects achieved by the present storage medium embodiment are also the same as those achieved by the above method embodiments.
While the preferred embodiments of the present invention have been illustrated and described, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (10)
1. A method for predicting liver cancer recurrence by combining imaging and deep learning is characterized by comprising the following steps:
acquiring a case image of a marked liver cancer target region;
inputting the case image of the marked liver cancer target area into a prediction model to obtain a prediction result;
the prediction model comprises a classification model, a three-dimensional convolution neural network model and a joint model, and is obtained through the following steps:
acquiring a case image sample of a marked liver cancer target region, and extracting an interested region image sample from the case image sample of the marked liver cancer target region;
extracting the image omics characteristics and three-dimensional small blocks of the image sample of the region of interest;
inputting the image omics characteristics into the classification model to obtain omics scores, and inputting the three-dimensional small blocks into the three-dimensional convolution neural network model to obtain network scores;
and inputting the omic score and the network score into the combined model to obtain a prediction score, determining a prediction result according to the prediction score, and determining the prediction model according to the accuracy of the prediction result.
2. The method for predicting liver cancer recurrence through the combination of imaging and deep learning as claimed in claim 1, wherein before extracting the imaging features of the region of interest image samples, the method further comprises:
and carrying out first standardization processing and three-dimensional reconstruction on the image sample of the region of interest.
3. The method for predicting liver cancer recurrence with imaging omics combined deep learning as set forth in claim 1, wherein prior to inputting said imaging omics features into said classification model to obtain an omics score, further comprising:
and performing dimension reduction processing on the image omics characteristics.
4. The method for predicting liver cancer recurrence through combination of imaging omics and deep learning as set forth in claim 1, wherein said three-dimensional patches are obtained by:
performing a second normalization process on the region-of-interest image sample;
extracting a plurality of three-dimensional small blocks for each image sample of the region of interest.
5. The method for predicting recurrence of liver cancer through combined imaging and deep learning as claimed in claim 1, wherein before inputting the three-dimensional patches into the three-dimensional convolutional neural network model to obtain network scores, the method further comprises:
and performing data enhancement on the three-dimensional small blocks.
6. The method for predicting liver cancer recurrence through combination of imaging omics and deep learning as set forth in claim 5, wherein the data enhancement of the three-dimensional patches comprises:
and turning, rotating, clipping, deforming, scaling, denoising, blurring, color transformation, erasing or filling the three-dimensional small blocks.
7. The method for predicting liver cancer recurrence through imaging omics combined deep learning as set forth in claim 1, wherein the three-dimensional convolutional neural network model comprises a first convolutional layer, a second convolutional layer, a first pooling layer, a third convolutional layer, a fourth convolutional layer, a first pooling layer, a first fully-connected layer, a second fully-connected layer, a third fully-connected layer, a fourth fully-connected layer and a classifier, which are connected in sequence, and the step of inputting the three-dimensional small block into the three-dimensional convolutional neural network model to obtain a network score specifically comprises the steps of:
and sequentially passing the three-dimensional small blocks through a first convolution layer, a second convolution layer, a first pooling layer, a third convolution layer, a fourth convolution layer, a first pooling layer, a first full-connection layer, a second full-connection layer, a third full-connection layer, a fourth full-connection layer and a classifier, and outputting a network score.
8. A system for predicting liver cancer recurrence by combining imaging and deep learning is characterized by comprising:
the acquisition module is used for acquiring a case image of the marked liver cancer target area;
the prediction module is used for inputting the case image of the marked liver cancer target area into a prediction model to obtain a prediction result;
the prediction model comprises a classification model, a three-dimensional convolution neural network model and a joint model, and is obtained through the following steps:
acquiring a case image sample of a marked liver cancer target region, and extracting an interested region image sample from the case image sample of the marked liver cancer target region;
extracting the image omics characteristics and three-dimensional small blocks of the image sample of the region of interest;
inputting the image omics characteristics into the classification model to obtain omics scores, and inputting the three-dimensional small blocks into the three-dimensional convolution neural network model to obtain network scores;
and inputting the omic score and the network score into the combined model to obtain a prediction score, determining a prediction result according to the prediction score, and determining the prediction model according to the accuracy of the prediction result.
9. A system for predicting liver cancer recurrence by combining imaging and deep learning is characterized by comprising:
at least one processor;
at least one memory for storing at least one program;
when executed by the at least one processor, cause the at least one processor to implement the method of any one of claims 1-7.
10. A storage medium having stored therein a program executable by a processor, wherein the program executable by the processor is adapted to perform the method of any one of claims 1-7 when executed by the processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110982206.9A CN113780274A (en) | 2021-08-25 | 2021-08-25 | Method, system and medium for predicting liver cancer recurrence by combining imaging omics and deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110982206.9A CN113780274A (en) | 2021-08-25 | 2021-08-25 | Method, system and medium for predicting liver cancer recurrence by combining imaging omics and deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113780274A true CN113780274A (en) | 2021-12-10 |
Family
ID=78839291
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110982206.9A Pending CN113780274A (en) | 2021-08-25 | 2021-08-25 | Method, system and medium for predicting liver cancer recurrence by combining imaging omics and deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113780274A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108364006A (en) * | 2018-01-17 | 2018-08-03 | 超凡影像科技股份有限公司 | Medical Images Classification device and its construction method based on multi-mode deep learning |
-
2021
- 2021-08-25 CN CN202110982206.9A patent/CN113780274A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108364006A (en) * | 2018-01-17 | 2018-08-03 | 超凡影像科技股份有限公司 | Medical Images Classification device and its construction method based on multi-mode deep learning |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
|---|---|---|
| CN109993726B (en) | Medical image detection method, device, equipment and storage medium | |
| CN110310287B (en) | Automatic organ-at-risk delineation method, equipment and storage medium based on neural network | |
| Bi et al. | Dual-path adversarial learning for fully convolutional network (FCN)-based medical image segmentation | |
| Nowak et al. | Fully automated segmentation of connective tissue compartments for CT-based body composition analysis: a deep learning approach | |
| Fang et al. | Globally guided progressive fusion network for 3D pancreas segmentation | |
| Pulagam et al. | Automated lung segmentation from HRCT scans with diffuse parenchymal lung diseases | |
| Albishri et al. | CU-Net: Cascaded U-Net model for automated liver and lesion segmentation and summarization | |
| Wardhana et al. | Toward reliable automatic liver and tumor segmentation using convolutional neural network based on 2.5 D models | |
| Dutande et al. | Deep residual separable convolutional neural network for lung tumor segmentation | |
| Antonelli et al. | Segmentation and reconstruction of the lung volume in CT images | |
| Hammon et al. | Model-based pancreas segmentation in portal venous phase contrast-enhanced CT images | |
| Aslam et al. | Liver-tumor detection using CNN ResUNet | |
| Tummala et al. | Liver tumor segmentation from computed tomography images using multiscale residual dilated encoder‐decoder network | |
| Hakim et al. | Microcalcification detection in mammography image using computer-aided detection based on convolutional neural network | |
| Hou et al. | Mass segmentation for whole mammograms via attentive multi-task learning framework | |
| CN113780274A (en) | Method, system and medium for predicting liver cancer recurrence by combining imaging omics and deep learning | |
| Harrison et al. | State-of-the-art of breast cancer diagnosis in medical images via convolutional neural networks (cnns) | |
| Li et al. | Automated classification of solitary pulmonary nodules using convolutional neural network based on transfer learning strategy | |
| CN114693671B (en) | Lung nodule semi-automatic segmentation method, device, equipment and medium based on deep learning | |
| Liu et al. | U2F-GAN: weakly supervised super-pixel segmentation in thyroid ultrasound images | |
| Yu et al. | CT segmentation of liver and tumors fused multi-scale features | |
| Mousavi Moghaddam et al. | Lung parenchyma segmentation from CT images with a fully automatic method | |
| Maghsoudi et al. | O-net: An overall convolutional network for segmentation tasks | |
| Hassan et al. | A deep learning model for mammography mass detection using mosaic and reconstructed multichannel images | |
| CN113362350A (en) | Segmentation method and device for cancer medical record image, terminal device and storage medium |
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: 20211210 |