CN111340825B - Method and system for generating mediastinum lymph node segmentation model - Google Patents
Method and system for generating mediastinum lymph node segmentation model Download PDFInfo
- Publication number
- CN111340825B CN111340825B CN202010148259.6A CN202010148259A CN111340825B CN 111340825 B CN111340825 B CN 111340825B CN 202010148259 A CN202010148259 A CN 202010148259A CN 111340825 B CN111340825 B CN 111340825B
- Authority
- CN
- China
- Prior art keywords
- image
- segmentation
- lymph node
- mediastinal lymph
- dimensional
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- 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
- G06T2207/10081—Computed x-ray tomography [CT]
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30061—Lung
-
- 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
Abstract
The invention provides a generation method and a generation system of a mediastinal lymph node segmentation model, which relate to the technical field of medical image processing and comprise the following steps: acquiring lung CT images of a plurality of thoracic surgery patients, and carrying out three-dimensional reconstruction on each lung CT image to obtain a three-dimensional image; respectively carrying out three-dimensional segmentation on each three-dimensional image to obtain a three-dimensional labeling image marked with a true mediastinal lymph node focus area; grouping the three-dimensional labeling images to obtain a training set, a testing set and a correction set; training the training set to obtain a mediastinal lymph node segmentation model; inputting the test set into the mediastinal lymph node segmentation model to obtain a corresponding segmentation image, and calculating the segmentation accuracy of the mediastinal lymph node segmentation model; the segmentation accuracy is smaller than the accuracy threshold, and the correction set corrects the mediastinal lymph node segmentation model; and if the segmentation accuracy is not less than the accuracy threshold, storing the mediastinal lymph node segmentation model. The invention effectively improves the accuracy of mediastinal lymph node segmentation, does not need manual intervention, and has strong practicability.
Description
Technical Field
The invention relates to the technical field of medical image processing, in particular to a generation method and a generation system of a mediastinal lymph node segmentation model.
Background
The incidence rate of lung cancer is high, the death rate is high, the survival rate of 5 years is low, and the lung cancer is the primary factor of global cancer death. Diffusion and metastasis occur in advanced stages of lung cancer, with mediastinal lymph node metastasis being a more common condition. Lung cancer has no obvious symptoms in the early stage of lymphatic metastasis, and lymphadenopathy can appear in the late stage. As the condition progresses, a plurality of lymphadenectasis occurs. After the occurrence of lymph node metastasis in lung cancer, most of them are not treated well, because the spread metastasis of cancer cells is very serious, and many new cancer lesions are formed throughout the whole body through the lymphatic system. Thus, accurate segmentation of lymph nodes is of great importance for surgical removal of lesions.
At present, medical imaging has become one of the most important means for lung cancer examination, treatment scheme selection, curative effect detection and the like, and according to medical imaging examination methods such as X-ray imaging, ultrasonic imaging, computed tomography (Computer Tomography, CT), magnetic resonance imaging (Magnetic Resonance Imaging, MRI), positron emission tomography (Positron Emission Tomography, PET) and the like, the method can non-invasively 'snoop' lesions in human bodies, examine tumors and monitor the change of the conditions of the tumors. However, in the case of conventional lymph node segmentation, a practitioner must perform manual segmentation by observing the image. The method requires complicated manual operation on a large amount of data by a professional doctor, and due to long-term and large-amount repeated work, errors of identification and edge segmentation can occur to the doctor, meanwhile, the accuracy and reliability of the segmentation result of the method are seriously dependent on the experience knowledge and professional quality of the doctor, and the accuracy of the result is limited, so that the research of an automatic lymph node segmentation system is necessary.
In recent years, machine learning and deep learning have been developed, and excellent performance has been exhibited in various fields. In the field of medical information analysis, machine learning and deep learning algorithms are widely applied in the fields of breast tumor classification, breast molybdenum target tumor detection and the like. The successful segmentation of the cell image by the U-network (U-Net) again demonstrates that deep learning can be well used for semantic segmentation of medical images. At present, the application of deep learning in clinical lung focus is mostly limited to lung nodules, however, the detection and segmentation of lung mediastinal lymph nodes have great significance for the formulation of doctor operation schemes, and no method for segmenting lung cancer mediastinal lymph nodes from CT images of lung cancer focus by using the deep learning is available at present.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a generation method of a mediastinal lymph node segmentation model, which specifically comprises the following steps:
step S1, acquiring lung CT images of a plurality of thoracic surgery patients, and carrying out three-dimensional reconstruction on each lung CT image to obtain a three-dimensional image corresponding to each thoracic surgery patient;
step S2, respectively carrying out three-dimensional segmentation on the three-dimensional images to obtain three-dimensional labeling images marked with the lesion areas of the true mediastinal lymph nodes;
s3, grouping the three-dimensional labeling images to obtain a training set, a testing set and a correction set;
step S4, training according to each three-dimensional labeling image in the training set to obtain a mediastinal lymph node segmentation model;
s5, inputting each three-dimensional labeling image in the test set into the mediastinal lymph node segmentation model to obtain corresponding segmentation images, and calculating the segmentation accuracy of the mediastinal lymph node segmentation model according to each segmentation image and the corresponding real mediastinal lymph node focus area;
step S6, comparing the segmentation accuracy with a preset accuracy threshold value:
if the segmentation accuracy is smaller than the accuracy threshold, turning to the step S7;
if the segmentation accuracy is not less than the accuracy threshold, turning to step S8;
step S7, correcting the mediastinal lymph node segmentation model according to each three-dimensional labeling image in the correction set, and returning to the step S6;
and S8, storing the mediastinal lymph node segmentation model to segment the mediastinal lymph nodes.
Preferably, the step S4 specifically includes:
step S41, a coordinate system is established on each three-dimensional annotation image in the training set, and overlapping sampling is carried out on each three-dimensional annotation image according to the coordinate system to obtain a plurality of image blocks;
the image blocks obtained by sampling each three-dimensional labeling image have the same quantity and the same preset size, and each image block has a central coordinate associated with the coordinate system;
step S42, inputting each image block into a pre-generated depth residual U-net segmentation model for feature learning according to each image block obtained by sampling each three-dimensional labeling image, and obtaining a sub-segmentation probability map corresponding to each image block;
each voxel point of the image block corresponding to the sub-segmentation probability map is a probability value of a mediastinal lymph node;
s43, restoring each sub-segmentation probability map in the coordinate system of each corresponding image block according to the central coordinates, and averaging the probability values of the coincident voxel points among the sub-segmentation probability maps to obtain a total segmentation probability map of the three-dimensional image;
step S44, comparing the probability value of each voxel point in the total segmentation probability map with a preset class probability threshold:
if the probability value is greater than the class probability threshold, setting the voxel value of the corresponding voxel point to be a first numerical value representing that the voxel point is a focus zone, and then turning to step S45;
if the probability value is not greater than the class probability threshold, setting the voxel value of the corresponding voxel point to a second value representing that the voxel point is not a focus zone, and then turning to step S45;
step S45, generating a mediastinal lymph node segmentation result of the three-dimensional labeling image according to the first numerical value and the second numerical value, calculating an error between the mediastinal lymph node segmentation result and the true mediastinal lymph node focus area, and adjusting parameters of the depth residual U-net segmentation model according to the error;
and step S46, repeating the steps S42 to S45 until training is finished, and obtaining a mediastinal lymph node segmentation model.
Preferably, in the step S41, the preset size is 24 pixels by 8 pixels.
Preferably, in the step S44, the first value is 1.
Preferably, in the step S44, the second value is 0.
Preferably, the step S5 specifically includes:
step S51, inputting each three-dimensional labeling image in the test set into the mediastinal lymph node segmentation model to obtain corresponding segmentation images;
step S52, calculating, according to each of the segmented images and the corresponding true mediastinal lymph node focal region, a coincidence rate between each of the segmented images and the corresponding true mediastinal lymph node focal region, and comparing the coincidence rate with a preset coincidence rate threshold value:
if the coincidence rate is not less than the coincidence rate threshold, adding the corresponding three-dimensional labeling image into a first image library, and then turning to step S53;
if the coincidence rate is smaller than the coincidence rate threshold, adding the corresponding three-dimensional labeling image into a second image library, and then turning to step S53;
step S53, respectively counting to obtain a first number of the three-dimensional labeling images in the first image library and a second number of the three-dimensional labeling images in the second image library, and calculating to obtain the segmentation accuracy of the mediastinal lymph node segmentation model according to the first number and the second number.
Preferably, in the step S53, the segmentation accuracy is calculated according to the following formula:
A r ＝N 1 /(N 1 +N 2 )
wherein, the liquid crystal display device comprises a liquid crystal display device,
A r representing the segmentation accuracy;
N 1 representing the first number;
N 2 representing the second number.
A generation system of a mediastinal lymph node segmentation model, to which the generation method of any one of the above mediastinal lymph node segmentation models is applied, the generation system specifically comprising:
the data acquisition module is used for acquiring lung CT images of a plurality of thoracic surgery patients, and carrying out three-dimensional reconstruction on each lung CT image to obtain a three-dimensional image corresponding to each thoracic surgery patient;
the data preprocessing module is connected with the data acquisition module and is used for respectively carrying out three-dimensional segmentation on the three-dimensional images to obtain three-dimensional labeling images marked with the true mediastinal lymph node focus areas;
the data grouping module is connected with the data preprocessing module and is used for grouping the three-dimensional annotation images to obtain a training set, a testing set and a correction set;
the model training module is connected with the data grouping module and is used for training according to each three-dimensional labeling image in the training set to obtain a mediastinal lymph node segmentation model;
the model evaluation module is respectively connected with the data grouping module and the model training module and is used for respectively inputting each three-dimensional labeling image in the test set into the mediastinal lymph node segmentation model to obtain corresponding segmentation images, and calculating the segmentation accuracy of the mediastinal lymph node segmentation model according to each segmentation image and the corresponding real mediastinal lymph node focus area;
the data comparison module is connected with the model evaluation module and is used for comparing the segmentation accuracy with a preset accuracy threshold, generating a first comparison result when the segmentation accuracy is not smaller than the accuracy threshold, and generating a second comparison result when the segmentation accuracy is smaller than the accuracy threshold;
the model correction module is respectively connected with the data grouping module and the data comparison module and is used for correcting the mediastinum lymph node segmentation model according to the first comparison result and each three-dimensional labeling image in the correction set;
and the model storage module is connected with the data comparison module and is used for storing the mediastinal lymph node segmentation model according to the second comparison result so as to segment the mediastinal lymph node.
Preferably, the model training module specifically includes:
the image sampling unit is used for establishing a coordinate system on each three-dimensional annotation image in the training set, and respectively sampling each three-dimensional annotation image according to the coordinate system to obtain a plurality of image blocks;
the image blocks obtained by sampling each three-dimensional labeling image have the same quantity and the same preset size, and each image block has a central coordinate associated with the coordinate system;
the feature learning unit is connected with the image sampling unit and is used for inputting each image block obtained by sampling each three-dimensional labeling image into a pre-generated depth residual U-net segmentation model to perform feature learning so as to obtain a sub-segmentation probability map corresponding to each image block;
each voxel point of the image block corresponding to the sub-segmentation probability map is a probability value of a mediastinal lymph node;
the image restoration unit is connected with the feature learning unit and is used for restoring each sub-segmentation probability map in the coordinate system where the corresponding image block is located according to the central coordinate, and averaging the probability values of the coincident voxel points among the sub-segmentation probability maps to obtain a total segmentation probability map of the three-dimensional image;
the probability comparison unit is connected with the image restoration unit and is used for comparing the probability value of each voxel point in the total segmentation probability map with a preset class probability threshold value, setting the voxel value of the corresponding voxel point as a first numerical value representing that the voxel point is a focus zone when the probability value is larger than the class probability threshold value, and
setting the voxel value of the corresponding voxel point to a second numerical value representing that the voxel point is not a focus zone when the probability value is not greater than the class probability threshold;
the data adjustment unit is connected with the data comparison unit and is used for generating a mediastinal lymph node segmentation result of the three-dimensional image according to the first numerical value and the second numerical value, calculating an error between the mediastinal lymph node segmentation result and the real mediastinal lymph node focus area, and adjusting parameters of the depth residual U-net segmentation model according to the error;
and the model generation unit is connected with the data adjustment unit and is used for obtaining a mediastinal lymph node segmentation model after training is finished.
Preferably, the model evaluation module specifically includes:
the image segmentation unit is used for inputting each three-dimensional labeling image in the test set into the mediastinal lymph node segmentation model to obtain corresponding segmentation images;
the image processing unit is connected with the image segmentation unit and is used for respectively calculating the coincidence rate between each segmented image and the corresponding true mediastinal lymph node focus area according to each segmented image and the corresponding true mediastinal lymph node focus area, adding the corresponding three-dimensional labeling image into a first image library when the coincidence rate is not smaller than a preset coincidence rate threshold value, and
when the coincidence rate is smaller than the coincidence rate threshold value, adding the corresponding three-dimensional annotation image into a second image library;
the data processing unit is connected with the image processing unit and is used for respectively counting and obtaining the first number of the three-dimensional labeling images in the first image library and the second number of the three-dimensional labeling images in the second image library, and calculating and obtaining the segmentation accuracy of the mediastinal lymph node segmentation model according to the first number and the second number.
The technical scheme has the following advantages or beneficial effects:
1) The accuracy and the specificity of the mediastinal lymph node segmentation model are effectively improved through the overlapping sampling of the three-dimensional images in the training set, manual intervention is not needed, and the practicability is high;
2) When the test result of the test set on the mediastinal lymph node segmentation model is not ideal, correcting the mediastinal lymph node segmentation model by adopting a correction set, so that the segmentation accuracy of the mediastinal lymph node segmentation model is further improved;
3) The automatic segmentation of the mediastinal lymph nodes can be realized through the mediastinal lymph node segmentation model, a doctor is helped to make further diagnosis, the condition is diagnosed in time, and an operation scheme is determined, so that the working intensity of the doctor can be relieved, and the best treatment opportunity of delaying the condition is avoided.
Drawings
FIG. 1 is a flow chart of a method for generating a mediastinal lymph node segmentation model in accordance with the preferred embodiment of the present invention;
FIG. 2 is a flow chart showing a training process of a mediastinal lymph node segmentation model according to the preferred embodiment of the invention;
FIG. 3 is a flowchart illustrating a method for calculating segmentation accuracy in a preferred embodiment of the present invention;
fig. 4 is a schematic structural diagram of a generation system of a mediastinal lymph node segmentation model according to a preferred embodiment of the present invention.
Detailed Description
The invention will now be described in detail with reference to the drawings and specific examples. The present invention is not limited to the embodiment, and other embodiments may fall within the scope of the present invention as long as they conform to the gist of the present invention.
In a preferred embodiment of the present invention, based on the above-mentioned problems existing in the prior art, a method for generating a mediastinal lymph node segmentation model is now provided, as shown in fig. 1, which specifically includes the following steps:
step S1, acquiring lung CT images of a plurality of thoracic surgery patients, and carrying out three-dimensional reconstruction on each lung CT image to obtain a three-dimensional image corresponding to each thoracic surgery patient;
step S2, respectively carrying out three-dimensional segmentation on each three-dimensional image to obtain a three-dimensional labeling image marked with a true mediastinal lymph node focus area;
step S3, grouping the three-dimensional labeling images to obtain a training set, a testing set and a correction set;
step S4, training according to each three-dimensional labeling image in the training set to obtain a mediastinal lymph node segmentation model;
s5, inputting each three-dimensional labeling image in the test set into a mediastinal lymph node segmentation model to obtain corresponding segmentation images, and calculating to obtain the segmentation accuracy of the mediastinal lymph node segmentation model according to each segmentation image and the corresponding real mediastinal lymph node focus area;
step S6, comparing the segmentation accuracy with a preset accuracy threshold value:
if the segmentation accuracy is smaller than the accuracy threshold, turning to the step S7;
if the segmentation accuracy is not less than the accuracy threshold, turning to step S8;
step S7, correcting the mediastinal lymph node segmentation model according to each three-dimensional labeling image in the correction set, and returning to the step S6;
and S8, storing the mediastinal lymph node segmentation model to segment the mediastinal lymph nodes.
Specifically, in this embodiment, 3700 sets of lung CT images including images of each stage of imaging dataset of a thoracic surgery patient are preferably acquired, each set of lung CT images includes a plurality of scan slice images, the scan slice images are two-dimensional images, and corresponding three-dimensional images are obtained by performing three-dimensional reconstruction on the scan slice images in each set of lung CT images. And then, carrying out three-dimensional segmentation on the mediastinal lymph nodes by using an image expert according to each three-dimensional image to obtain a three-dimensional labeling image marked with a true mediastinal lymph node focus area. Preferably, randomly dividing 3700 three-dimensional labeling images into a training set, a testing set and a correcting set, wherein the training set comprises 2000 three-dimensional labeling images and is used for training to obtain a mediastinal lymph node segmentation model; the test set comprises 1200 three-dimensional labeling images and is used for testing and evaluating the mediastinal lymph node segmentation model obtained through training, and the segmentation accuracy of the mediastinal lymph node segmentation model is used as an evaluation standard; the correction set comprises 500 three-dimensional labeling images and is used for correcting the mediastinal lymph node segmentation model when the segmentation accuracy rate does not reach the preset standard, and the obtained mediastinal lymph node segmentation model is used for segmentation of the mediastinal lymph nodes.
In a preferred embodiment of the present invention, as shown in fig. 2, step S4 specifically includes:
step S41, a coordinate system is established on each three-dimensional labeling image in the training set, and overlapping sampling is carried out on each three-dimensional labeling image according to the coordinate system to obtain a plurality of image blocks;
each three-dimensional labeling image is sampled to obtain image blocks with the same quantity and the same preset size, and each image block is provided with a center coordinate associated with a coordinate system;
step S42, inputting each image block into a pre-generated depth residual U-net segmentation model for feature learning aiming at each image block obtained by sampling each three-dimensional labeling image, and obtaining a sub-segmentation probability map corresponding to each image block;
each voxel point of the image block which is included in the sub-segmentation probability map is a probability value of the mediastinal lymph node;
s43, restoring each sub-segmentation probability map in a coordinate system where each corresponding image block is located according to the center coordinates, and averaging probability values of coincident voxel points among the sub-segmentation probability maps to obtain a total segmentation probability map of the three-dimensional image;
step S44, comparing the probability value of each voxel point in the total segmentation probability map with a preset class probability threshold value:
if the probability value is greater than the class probability threshold, setting the voxel value of the corresponding voxel point as a first numerical value representing that the voxel point is a focus zone, and then turning to step S45;
if the probability value is not greater than the class probability threshold, setting the voxel value of the corresponding voxel point as a second numerical value representing that the voxel point is not a focus zone, and then turning to step S45;
step S45, generating a mediastinal lymph node segmentation result of the three-dimensional labeling image according to the first numerical value and the second numerical value, calculating an error between the mediastinal lymph node segmentation result and a true mediastinal lymph node focus area, and adjusting parameters of a depth residual U-net segmentation model according to the error;
step S46, repeating the steps S42 to S45 until training is finished, and obtaining the mediastinal lymph node segmentation model.
Specifically, in this embodiment, the mediastinal lymph node segmentation model includes three processing procedures, where the first processing procedure is a procedure of sampling a three-dimensional labeling image:
in this embodiment, the three-dimensional labeling image is preferably sampled into small blocks as input data of the next processing procedure in a sampling manner with the true mediastinal lymph node lesion area as the center. Because the sizes of the true mediastinal lymph node focus areas in each three-dimensional labeling image are different, in order to enable the image blocks obtained by sampling each three-dimensional labeling image to have the same number of preset sizes, in the sampling process, different sampling intervals are preferably adopted for the three-dimensional labeling images with the true mediastinal lymph node focus areas with different sizes, and larger sampling intervals are preferably adopted for the three-dimensional labeling images with the larger true mediastinal lymph node focus areas; three-dimensional labeling images with smaller true mediastinal lymph node lesion areas employ smaller sampling intervals.
The second processing procedure is a rough segmentation process of the image block, and the depth residual U-net segmentation model is adopted to carry out rough segmentation of the image block:
the network structure of the depth residual U-net segmentation model comprises a compression process and an expansion process, wherein in the compression process, an input image block is firstly processed by a convolution layer, a batch normalization layer, an activation layer, a convolution layer and an addition layer to obtain a first feature map; the first feature map is subjected to downsampling feature compression to obtain a second feature map; the second feature map is subjected to downsampling feature compression to obtain a third feature map; the third feature map is subjected to downsampling feature compression to obtain a fourth feature map; in the expansion process, the fourth feature map is subjected to up-sampling to obtain a fifth feature map, the fifth feature map and the third feature map are subjected to feature fusion, then up-sampling is performed to obtain a sixth feature map, the sixth feature map and the second feature map are subjected to feature fusion, then up-sampling is performed to obtain a seventh feature map, the seventh feature map and the first feature map are subjected to feature fusion, and then convolution processing is performed, and then the seventh feature map and the third feature map are subjected to activating multiplication with Sigmoid to obtain a sub-segmentation probability map corresponding to an image block;
the third process is a sub-division process of image blocks:
and after the sub-segmentation probability map is obtained, restoring each image block to the original position according to the center coordinates of the image block, and averaging probability values of coincident voxel points among the sub-segmentation probability maps to obtain a total segmentation probability map of the three-dimensional image. Because the image block segmentation results are obtained by taking probability average on the overlapping voxel points, and the class probability is used as the result of the voxel values of the focus area, false positive is improved, and errors are generated. In order to optimize the segmentation result, removing the false focus points, further reducing false positives, setting a class probability threshold, setting a voxel value at a position larger than the class probability threshold to be a first value, preferably 1, and setting a voxel value at a position not larger than the class probability threshold to be a second value, preferably 0. Voxel values with low class probability appear in the rough segmentation result, but the region does not belong to a focus region, and the region is set as a healthy region by setting a class probability threshold value, so that a precise segmentation result is obtained, and the establishment of a mediastinal lymph node segmentation model is realized.
In the preferred embodiment of the present invention, in step S41, the preset size is 24 pixels by 8 pixels.
In a preferred embodiment of the present invention, in step S44, the first value is 1.
In a preferred embodiment of the present invention, in step S44, the second value is 0.
In a preferred embodiment of the present invention, as shown in fig. 3, step S5 specifically includes:
step S51, respectively inputting each three-dimensional labeling image in the test set into a mediastinal lymph node segmentation model to obtain corresponding segmentation images;
step S52, according to each segmented image and the corresponding real mediastinal lymph node focus area, the coincidence rate between each segmented image and the corresponding real mediastinal lymph node focus area is calculated, and the coincidence rate is compared with a preset coincidence rate threshold value:
if the coincidence rate is not less than the coincidence rate threshold, adding the corresponding three-dimensional labeling image into a first image library, and then turning to step S53;
if the coincidence rate is smaller than the coincidence rate threshold, adding the corresponding three-dimensional labeling image into a second image library, and then turning to step S53;
and step S53, respectively counting to obtain a first number of three-dimensional labeling images in the first image library and a second number of three-dimensional labeling images in the second image library, and calculating to obtain the segmentation accuracy of the mediastinal lymph node segmentation model according to the first number and the second number.
In a preferred embodiment of the present invention, in step S53, the segmentation accuracy is calculated according to the following formula:
A r ＝N 1 /(N 1 +N 2 )
wherein, the liquid crystal display device comprises a liquid crystal display device,
A r representing segmentation accuracy;
N 1 representing a first number;
N 2 representing a second number.
A generation system of a mediastinal lymph node segmentation model, which applies the generation method of the mediastinal lymph node segmentation model of any one of the above, as shown in fig. 4, specifically includes:
the data acquisition module 1 is used for acquiring lung CT images of a plurality of thoracic surgery patients, and carrying out three-dimensional reconstruction on each lung CT image to obtain a corresponding three-dimensional image of each thoracic surgery patient;
the data preprocessing module 2 is connected with the data acquisition module 1 and is used for respectively carrying out three-dimensional segmentation on each three-dimensional image to obtain a three-dimensional labeling image marked with a true mediastinal lymph node focus area;
the data grouping module 3 is connected with the data preprocessing module 2 and is used for grouping all the three-dimensional annotation images to obtain a training set, a testing set and a correction set;
the model training module 4 is connected with the data grouping module 4 and is used for training according to each three-dimensional labeling image in the training set to obtain a mediastinal lymph node segmentation model;
the model evaluation module 5 is respectively connected with the data grouping module 3 and the model training module 4 and is used for respectively inputting each three-dimensional labeling image in the test set into the mediastinal lymph node segmentation model to obtain corresponding segmentation images, and calculating the segmentation accuracy of the mediastinal lymph node segmentation model according to each segmentation image and the corresponding real mediastinal lymph node focus area;
the data comparison module 6 is connected with the model evaluation module 5 and is used for comparing the segmentation accuracy with a preset accuracy threshold, generating a first comparison result when the segmentation accuracy is not smaller than the accuracy threshold, and generating a second comparison result when the segmentation accuracy is smaller than the accuracy threshold;
the model correction module 7 is respectively connected with the data grouping module 3 and the data comparison module 6 and is used for correcting the mediastinum lymph node segmentation model according to the first comparison result and each three-dimensional labeling image in the correction set;
the model storage module 8 is connected with the data comparison module 7 and is used for storing the mediastinal lymph node segmentation model according to the second comparison result so as to segment the mediastinal lymph node.
In a preferred embodiment of the present invention, as shown in fig. 4, the model training module 4 specifically includes:
the image sampling unit 41 is configured to establish a coordinate system on each three-dimensional labeling image in the training set, and sample each three-dimensional labeling image according to the coordinate system to obtain a plurality of image blocks;
each three-dimensional labeling image is sampled to obtain image blocks with the same quantity and the same preset size, and each image block is provided with a center coordinate associated with a coordinate system;
the feature learning unit 42 is connected with the image sampling unit 41, and is configured to input each image block obtained by sampling each three-dimensional labeling image into a depth residual error U-net segmentation model generated in advance to perform feature learning, so as to obtain a sub-segmentation probability map corresponding to each image block;
each voxel point of the image block which is included in the sub-segmentation probability map is a probability value of the mediastinal lymph node;
an image restoration unit 43, connected to the feature learning unit 42, configured to restore each sub-segmentation probability map to a coordinate system where each corresponding image block is located according to the central coordinate, and average probability values of overlapping voxel points between each sub-segmentation probability map to obtain a total segmentation probability map of the three-dimensional image;
a probability comparison unit 44 connected to the image restoration unit 43 for comparing the probability value of each voxel point in the total segmentation probability map with a preset class probability threshold, setting the voxel value of the corresponding voxel point as a first value representing that the voxel point is a focus zone when the probability value is greater than the class probability threshold, and
when the probability value is not greater than the class probability threshold, setting the voxel value of the corresponding voxel point as a second numerical value which characterizes that the voxel point is not a focus zone;
the data adjustment unit 45 is connected with the data comparison unit 44, and is used for generating a mediastinal lymph node segmentation result of the three-dimensional image according to the first numerical value and the second numerical value, calculating an error between the mediastinal lymph node segmentation result and a true mediastinal lymph node lesion area, and adjusting parameters of the depth residual U-net segmentation model according to the error;
the model generation unit 46 is connected to the data adjustment unit 45, and is configured to obtain a mediastinal lymph node segmentation model at the end of training.
In a preferred embodiment of the present invention, as shown in fig. 4, the model evaluation module 5 specifically includes:
an image segmentation unit 51, configured to input each three-dimensional labeling image in the test set into a mediastinal lymph node segmentation model to obtain a corresponding segmented image;
an image processing unit 52 connected to the image segmentation unit 51 for calculating the coincidence rate between each segmented image and the corresponding true mediastinal lymph node focal region according to each segmented image and the corresponding true mediastinal lymph node focal region, respectively, and adding the corresponding three-dimensional labeling image into a first image library when the coincidence rate is not less than a preset coincidence rate threshold, and
when the coincidence rate is smaller than the coincidence rate threshold value, adding the corresponding three-dimensional annotation image into a second image library;
the data processing unit 53 is connected to the image processing unit 52, and is configured to count a first number of three-dimensional labeling images in the first image library and a second number of three-dimensional labeling images in the second image library, and obtain a segmentation accuracy of the mediastinal lymph node segmentation model according to the first number and the second number.
The foregoing description is only illustrative of the preferred embodiments of the present invention and is not to be construed as limiting the scope of the invention, and it will be appreciated by those skilled in the art that equivalent substitutions and obvious variations may be made using the description and drawings, and are intended to be included within the scope of the present invention.
Claims (9)
1. The generation method of the mediastinal lymph node segmentation model is characterized by comprising the following steps of:
step S1, acquiring lung CT images of a plurality of thoracic surgery patients, and carrying out three-dimensional reconstruction on each lung CT image to obtain a three-dimensional image corresponding to each thoracic surgery patient;
step S2, respectively carrying out three-dimensional segmentation on the three-dimensional images to obtain three-dimensional labeling images marked with the lesion areas of the true mediastinal lymph nodes;
s3, grouping the three-dimensional labeling images to obtain a training set, a testing set and a correction set;
step S4, training according to each three-dimensional labeling image in the training set to obtain a mediastinal lymph node segmentation model;
s5, inputting each three-dimensional labeling image in the test set into the mediastinal lymph node segmentation model to obtain corresponding segmentation images, and calculating the segmentation accuracy of the mediastinal lymph node segmentation model according to each segmentation image and the corresponding real mediastinal lymph node focus area;
step S6, comparing the segmentation accuracy with a preset accuracy threshold value:
if the segmentation accuracy is smaller than the accuracy threshold, turning to the step S7;
if the segmentation accuracy is not less than the accuracy threshold, turning to step S8;
step S7, correcting the mediastinal lymph node segmentation model according to each three-dimensional labeling image in the correction set, and returning to the step S6;
step S8, the mediastinal lymph node segmentation model is stored so as to segment the mediastinal lymph nodes;
the step S4 specifically includes:
step S41, a coordinate system is established on each three-dimensional annotation image in the training set, and overlapping sampling is carried out on each three-dimensional annotation image according to the coordinate system to obtain a plurality of image blocks;
the image blocks obtained by sampling each three-dimensional labeling image have the same quantity and the same preset size, and each image block has a central coordinate associated with the coordinate system;
step S42, inputting each image block into a pre-generated depth residual U-net segmentation model for feature learning according to each image block obtained by sampling each three-dimensional labeling image, and obtaining a sub-segmentation probability map corresponding to each image block;
each voxel point of the image block corresponding to the sub-segmentation probability map is a probability value of a mediastinal lymph node;
s43, restoring each sub-segmentation probability map in the coordinate system of each corresponding image block according to the central coordinates, and averaging the probability values of the coincident voxel points among the sub-segmentation probability maps to obtain a total segmentation probability map of the three-dimensional image;
step S44, comparing the probability value of each voxel point in the total segmentation probability map with a preset class probability threshold:
if the probability value is greater than the class probability threshold, setting the voxel value of the corresponding voxel point to be a first numerical value representing that the voxel point is a focus zone, and then turning to step S45;
if the probability value is not greater than the class probability threshold, setting the voxel value of the corresponding voxel point to a second value representing that the voxel point is not a focus zone, and then turning to step S45;
step S45, generating a mediastinal lymph node segmentation result of the three-dimensional labeling image according to the first numerical value and the second numerical value, calculating an error between the mediastinal lymph node segmentation result and the true mediastinal lymph node focus area, and adjusting parameters of the depth residual U-net segmentation model according to the error;
and step S46, repeating the steps S42 to S45 until training is finished, and obtaining a mediastinal lymph node segmentation model.
2. The method according to claim 1, wherein in the step S41, the preset size is 24 pixels by 8 pixels.
3. The method according to claim 1, wherein in the step S44, the first value is 1.
4. The method according to claim 1, wherein in the step S44, the second value is 0.
5. The method for generating a mediastinal lymph node segmentation model according to claim 1, wherein the step S5 specifically comprises:
step S51, inputting each three-dimensional labeling image in the test set into the mediastinal lymph node segmentation model to obtain corresponding segmentation images;
step S52, calculating, according to each of the segmented images and the corresponding true mediastinal lymph node focal region, a coincidence rate between each of the segmented images and the corresponding true mediastinal lymph node focal region, and comparing the coincidence rate with a preset coincidence rate threshold value:
if the coincidence rate is not less than the coincidence rate threshold, adding the corresponding three-dimensional labeling image into a first image library, and then turning to step S53;
if the coincidence rate is smaller than the coincidence rate threshold, adding the corresponding three-dimensional labeling image into a second image library, and then turning to step S53;
step S53, respectively counting to obtain a first number of the three-dimensional labeling images in the first image library and a second number of the three-dimensional labeling images in the second image library, and calculating to obtain the segmentation accuracy of the mediastinal lymph node segmentation model according to the first number and the second number.
6. The method according to claim 5, wherein in the step S53, the segmentation accuracy is calculated according to the following formula:
A r ＝N 1 /(N 1 +N 2 )
wherein, the liquid crystal display device comprises a liquid crystal display device,
A r representing the segmentation accuracy;
N 1 representing the first number;
N 2 representing the second number.
7. A generation system of a mediastinal lymph node segmentation model, characterized in that a generation method of a mediastinal lymph node segmentation model according to any of claims 1-6 is applied, the generation system specifically comprising:
the data acquisition module is used for acquiring lung CT images of a plurality of thoracic surgery patients, and carrying out three-dimensional reconstruction on each lung CT image to obtain a three-dimensional image corresponding to each thoracic surgery patient;
the data preprocessing module is connected with the data acquisition module and is used for respectively carrying out three-dimensional segmentation on the three-dimensional images to obtain three-dimensional labeling images marked with the true mediastinal lymph node focus areas;
the data grouping module is connected with the data preprocessing module and is used for grouping the three-dimensional annotation images to obtain a training set, a testing set and a correction set;
the model training module is connected with the data grouping module and is used for training according to each three-dimensional labeling image in the training set to obtain a mediastinal lymph node segmentation model;
the model evaluation module is respectively connected with the data grouping module and the model training module and is used for respectively inputting each three-dimensional labeling image in the test set into the mediastinal lymph node segmentation model to obtain corresponding segmentation images, and calculating the segmentation accuracy of the mediastinal lymph node segmentation model according to each segmentation image and the corresponding real mediastinal lymph node focus area;
the data comparison module is connected with the model evaluation module and is used for comparing the segmentation accuracy with a preset accuracy threshold, generating a first comparison result when the segmentation accuracy is not smaller than the accuracy threshold, and generating a second comparison result when the segmentation accuracy is smaller than the accuracy threshold;
the model correction module is respectively connected with the data grouping module and the data comparison module and is used for correcting the mediastinum lymph node segmentation model according to the first comparison result and each three-dimensional labeling image in the correction set;
and the model storage module is connected with the data comparison module and is used for storing the mediastinal lymph node segmentation model according to the second comparison result so as to segment the mediastinal lymph node.
8. The system for generating a mediastinal lymph node segmentation model of claim 7, wherein the model training module specifically comprises:
the image sampling unit is used for establishing a coordinate system on each three-dimensional annotation image in the training set, and respectively sampling each three-dimensional annotation image according to the coordinate system to obtain a plurality of image blocks;
the image blocks obtained by sampling each three-dimensional labeling image have the same quantity and the same preset size, and each image block has a central coordinate associated with the coordinate system;
the feature learning unit is connected with the image sampling unit and is used for inputting each image block obtained by sampling each three-dimensional labeling image into a pre-generated depth residual U-net segmentation model to perform feature learning so as to obtain a sub-segmentation probability map corresponding to each image block;
each voxel point of the image block corresponding to the sub-segmentation probability map is a probability value of a mediastinal lymph node;
the image restoration unit is connected with the feature learning unit and is used for restoring each sub-segmentation probability map in the coordinate system where the corresponding image block is located according to the central coordinate, and averaging the probability values of the coincident voxel points among the sub-segmentation probability maps to obtain a total segmentation probability map of the three-dimensional image;
the probability comparison unit is connected with the image restoration unit and is used for comparing the probability value of each voxel point in the total segmentation probability map with a preset class probability threshold value, setting the voxel value of the corresponding voxel point as a first numerical value representing that the voxel point is a focus zone when the probability value is larger than the class probability threshold value, and
setting the voxel value of the corresponding voxel point to a second numerical value representing that the voxel point is not a focus zone when the probability value is not greater than the class probability threshold;
the data adjustment unit is connected with the data comparison unit and is used for generating a mediastinal lymph node segmentation result of the three-dimensional image according to the first numerical value and the second numerical value, calculating an error between the mediastinal lymph node segmentation result and the real mediastinal lymph node focus area, and adjusting parameters of the depth residual U-net segmentation model according to the error;
and the model generation unit is connected with the data adjustment unit and is used for obtaining a mediastinal lymph node segmentation model after training is finished.
9. The system for generating a mediastinal lymph node segmentation model according to claim 7, wherein the model evaluation module specifically comprises:
the image segmentation unit is used for inputting each three-dimensional labeling image in the test set into the mediastinal lymph node segmentation model to obtain corresponding segmentation images;
the image processing unit is connected with the image segmentation unit and is used for respectively calculating the coincidence rate between each segmented image and the corresponding true mediastinal lymph node focus area according to each segmented image and the corresponding true mediastinal lymph node focus area, adding the corresponding three-dimensional labeling image into a first image library when the coincidence rate is not smaller than a preset coincidence rate threshold value, and
when the coincidence rate is smaller than the coincidence rate threshold value, adding the corresponding three-dimensional annotation image into a second image library;
the data processing unit is connected with the image processing unit and is used for respectively counting and obtaining the first number of the three-dimensional labeling images in the first image library and the second number of the three-dimensional labeling images in the second image library, and calculating and obtaining the segmentation accuracy of the mediastinal lymph node segmentation model according to the first number and the second number.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010148259.6A CN111340825B (en) | 2020-03-05 | 2020-03-05 | Method and system for generating mediastinum lymph node segmentation model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010148259.6A CN111340825B (en) | 2020-03-05 | 2020-03-05 | Method and system for generating mediastinum lymph node segmentation model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111340825A CN111340825A (en) | 2020-06-26 |
| CN111340825B true CN111340825B (en) | 2023-05-09 |
Family
ID=71185914
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010148259.6A Active CN111340825B (en) | 2020-03-05 | 2020-03-05 | Method and system for generating mediastinum lymph node segmentation model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111340825B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112258500B (en) * | 2020-11-10 | 2023-08-04 | 北京深睿博联科技有限责任公司 | Lymph node partition method, apparatus, device, and computer-readable storage medium |
| CN113450381B (en) * | 2021-06-16 | 2022-10-18 | 上海深至信息科技有限公司 | System and method for evaluating accuracy of image segmentation model |
| CN113408633B (en) * | 2021-06-29 | 2023-04-18 | 北京百度网讯科技有限公司 | Method, apparatus, device and storage medium for outputting information |
| CN114820535B (en) * | 2022-05-05 | 2023-09-12 | 深圳市铱硙医疗科技有限公司 | Image detection method and device for aneurysm, computer equipment and storage medium |
| CN115272206B (en) * | 2022-07-18 | 2023-07-04 | 深圳市医未医疗科技有限公司 | Medical image processing method, medical image processing device, computer equipment and storage medium |
| CN115131345B (en) * | 2022-08-29 | 2023-02-03 | 杭州堃博生物科技有限公司 | CT image-based focus detection method and device and computer-readable storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107016665A (en) * | 2017-02-16 | 2017-08-04 | 浙江大学 | A kind of CT pulmonary nodule detection methods based on depth convolutional neural networks |
| CN109685811A (en) * | 2018-12-24 | 2019-04-26 | 北京大学第三医院 | PET/CT hypermetabolism lymph node dividing method based on dual path U-net convolutional neural networks |
| CN109949276A (en) * | 2019-02-28 | 2019-06-28 | 华中科技大学 | A kind of lymph node detection method in improvement SegNet segmentation network |
| CN110310289A (en) * | 2019-06-17 | 2019-10-08 | 北京交通大学 | Lung tissue's image partition method based on deep learning |
| CN110415253A (en) * | 2019-05-06 | 2019-11-05 | 南京大学 | A kind of point Interactive medical image dividing method based on deep neural network |
| CN110689525A (en) * | 2019-09-09 | 2020-01-14 | 上海中医药大学附属龙华医院 | Method and device for recognizing lymph nodes based on neural network |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107492071B (en) * | 2017-08-17 | 2021-04-27 | 京东方科技集团股份有限公司 | Medical image processing method and equipment |
| CA3041140C (en) * | 2018-04-26 | 2021-12-14 | NeuralSeg Ltd. | Systems and methods for segmenting an image |
-
2020
- 2020-03-05 CN CN202010148259.6A patent/CN111340825B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107016665A (en) * | 2017-02-16 | 2017-08-04 | 浙江大学 | A kind of CT pulmonary nodule detection methods based on depth convolutional neural networks |
| CN109685811A (en) * | 2018-12-24 | 2019-04-26 | 北京大学第三医院 | PET/CT hypermetabolism lymph node dividing method based on dual path U-net convolutional neural networks |
| CN109949276A (en) * | 2019-02-28 | 2019-06-28 | 华中科技大学 | A kind of lymph node detection method in improvement SegNet segmentation network |
| CN110415253A (en) * | 2019-05-06 | 2019-11-05 | 南京大学 | A kind of point Interactive medical image dividing method based on deep neural network |
| CN110310289A (en) * | 2019-06-17 | 2019-10-08 | 北京交通大学 | Lung tissue's image partition method based on deep learning |
| CN110689525A (en) * | 2019-09-09 | 2020-01-14 | 上海中医药大学附属龙华医院 | Method and device for recognizing lymph nodes based on neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111340825A (en) | 2020-06-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111340825B (en) | Method and system for generating mediastinum lymph node segmentation model | |
| CN111080584B (en) | Quality control method for medical image, computer device and readable storage medium | |
| US9218661B2 (en) | Image analysis for specific objects | |
| CN112508965B (en) | Automatic outline sketching system for normal organs in medical image | |
| JP2023507109A (en) | Automated tumor identification and segmentation from medical images | |
| CN112885453A (en) | Method and system for identifying pathological changes in subsequent medical images | |
| KR102328229B1 (en) | AI based tumor detection and diagnostic method using 3D medical image | |
| US8520923B2 (en) | Reporting organ volume for a medical digital image | |
| CN111553892A (en) | Lung nodule segmentation calculation method, device and system based on deep learning | |
| Moreau et al. | Deep learning approaches for bone and bone lesion segmentation on 18FDG PET/CT imaging in the context of metastatic breast cancer | |
| CN114332132A (en) | Image segmentation method and device and computer equipment | |
| CN115100067A (en) | Image reconstruction method and device | |
| CN111311626A (en) | Skull fracture automatic detection method based on CT image and electronic medium | |
| CN110738633B (en) | Three-dimensional image processing method and related equipment for organism tissues | |
| CN110992312B (en) | Medical image processing method, medical image processing device, storage medium and computer equipment | |
| CN112529900A (en) | Method, device, terminal and storage medium for matching ROI in mammary gland image | |
| Dandıl et al. | A Mask R-CNN based Approach for Automatic Lung Segmentation in Computed Tomography Scans | |
| CN114820483A (en) | Image detection method and device and computer equipment | |
| US11817204B2 (en) | Specialized computer-aided diagnosis and disease characterization with a multi-focal ensemble of convolutional neural networks | |
| Malinda et al. | Lumbar vertebrae synthetic segmentation in computed tomography images using hybrid deep generative adversarial networks | |
| CN110738664B (en) | Image positioning method and device, computer equipment and storage medium | |
| CN109447974B (en) | Volume data processing method, volume data processing apparatus, image processing workstation, and readable storage medium | |
| Rodriguez Ruiz | Artificial intelligence and tomosynthesis for breast cancer detection | |
| Wormanns et al. | Clinical evaluation of the reproducibility of volume measurements of pulmonary nodules | |
| US11615881B2 (en) | Liver cancer detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |