CN113160153A - Lung nodule screening method and system based on deep learning technology - Google Patents
Lung nodule screening method and system based on deep learning technology Download PDFInfo
- Publication number
- CN113160153A CN113160153A CN202110368194.0A CN202110368194A CN113160153A CN 113160153 A CN113160153 A CN 113160153A CN 202110368194 A CN202110368194 A CN 202110368194A CN 113160153 A CN113160153 A CN 113160153A
- Authority
- CN
- China
- Prior art keywords
- lung
- image
- module
- screening
- nodule
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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
- G06T7/155—Segmentation; Edge detection involving morphological operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/20—ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
-
- 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
- G06T2207/20036—Morphological image processing
-
- 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
- G06T2207/30064—Lung nodule
Abstract
The invention belongs to the technical field of pulmonary nodule screening, and discloses a pulmonary nodule screening method and a pulmonary nodule screening system based on a deep learning technology, wherein the pulmonary nodule screening system based on the deep learning technology comprises: the system comprises an information acquisition module, a lung image acquisition module, an image processing module, a central control module, an image analysis module, a lung tissue segmentation module, a feature extraction module, a screening model construction module, a model training module, an image segmentation module, a lung nodule qualitative module, a cloud storage module and an updating display module. According to the method, the nodules on the lung wall can be prevented from being missed to be detected through the concave-convex performance of the 2D space according to the segmentation contour; the classification of various types of nodules can be realized through a focus classification method based on deep learning, the local three-dimensional information of the focus of the CT sequence image is fully utilized, whether the nodules are lung nodules or not is effectively distinguished, the types of the nodules are identified, and doctors are better assisted to improve the diagnosis accuracy; the screening efficiency of the pulmonary nodules is improved, and the false positive nodules are reduced.
Description
Technical Field
The invention belongs to the technical field of pulmonary nodule screening, and particularly relates to a pulmonary nodule screening method and a pulmonary nodule screening system based on a deep learning technology.
Background
At present, lung cancer is a common malignant tumor, so that early detection and treatment are important, and judging whether lung nodules exist in the lung is an effective method for early prevention of the lung cancer. At present, a large number of CT images are screened by doctors for the existence of lung nodules, so that the workload of the doctors is increased, and missed diagnosis or misdiagnosis is easily caused. The computer-aided diagnosis system can process the lung CT image by using a machine learning method, and finally screen the lung nodules in the lung CT image. Therefore, the reading pressure of doctors can be greatly reduced, the opinion for judging the lung nodules can be provided for the doctors, and the method has important significance for the diagnosis of the lung cancer.
In the current automatic identification and detection technology of lung nodules in CT images, a convolutional neural network is mainly used for extracting the features of the images so as to detect the lung nodules in the images. However, the lung region contains many nodules-like physiological structures, such as pulmonary blood vessels, and there are many sizes, shapes and types of lung nodules, so that false positives are high in the lung nodule detection results.
Through the above analysis, the problems and defects of the prior art are as follows: in the current automatic identification and detection technology of lung nodules in CT images, the segmentation of lung structures is not carried out when the lung nodules in the images are detected, and false positives in lung nodule detection results are high.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a pulmonary nodule screening method and a pulmonary nodule screening system based on a deep learning technology.
The invention is realized in such a way that a pulmonary nodule screening method based on a deep learning technology comprises the following steps:
setting a case information acquisition parameter at a terminal, and sending an information acquisition request carrying the case information acquisition parameter to a server through an information acquisition module; the server receives an information acquisition request sent by the terminal, acquires case information acquisition parameters in the information acquisition request, and acquires a target case information set in the database based on the case information acquisition parameters; the server sets case information feedback parameters and sends a target case information set and the case information feedback parameters to the terminal; the terminal receives and stores a target case information set and case information feedback parameters, sets case information acquisition parameters based on the case information feedback parameters through an information acquisition program, and sends a case information acquisition request carrying the case information acquisition parameters to a server to obtain case information;
acquiring a lung image of the patient by using a CT machine through a lung image acquisition module to obtain a three-dimensional lung image; acquiring a lung three-dimensional image to be processed through an image processing module, and determining a state type of the image to be processed; searching a corresponding sampling model according to the state type by using an image processing program; carrying out nonlinear sampling on the image to be processed according to the sampling model to obtain a sampling point, and carrying out normalization processing on the lung three-dimensional image to be processed according to the sampling point;
step three, the server terminal adopts a template matching algorithm based on shape characteristics and a detection algorithm based on SIFT characteristics for detection, calls an image processing operator packaged in an OpenCV visual class library through an image analysis module, programs corresponding C + + codes, and realizes the function of the algorithm by utilizing each control button in an application software interface;
step four, image preprocessing is carried out by utilizing an image analysis program, image filtering processing, target sub-image acquisition and histogram equalization processing are completed by the image preprocessing, useless information contained in the image data is deleted, and image information of the area to be treated is obtained; image segmentation and automatic quantitative analysis are carried out to obtain the external contour characteristics of the target, on the basis of image segmentation, the number of pixels contained in the longest and widest distances of a target area is respectively obtained, and the pixels are converted into the length and width values of an entity according to a scale coefficient to complete three-dimensional visual angle conversion; displaying tissues of different parts according to interactive region selection, realizing analysis of the processed three-dimensional lung image, and generating a lung three-dimensional image analysis report;
fifthly, segmenting the lung region in the lung three-dimensional image by utilizing lung tissue segmentation through a lung tissue segmentation module to obtain a lung image; comparing the case result with the lung image by using a feature extraction program through a feature extraction module to obtain and label the feature points of the pathological part; constructing a pulmonary nodule screening model according to the labeled characteristic points by using a screening model construction module and a pulmonary nodule screening model construction program;
step six, training the pulmonary nodule screening model by using a model training program through a model training module to obtain a trained pulmonary nodule screening model; the image segmentation module is used for segmenting the lung three-dimensional image by using an image segmentation program to obtain a small-size image; screening the small-size image by using a nodule qualitative program through a pulmonary nodule qualitative module according to the trained pulmonary nodule screening model to obtain a pulmonary nodule qualitative result;
step seven, storing the acquired case information, the lung three-dimensional image analysis result, the lung image, the lesion part feature point, the lung nodule screening model, the small-size image and the lung nodule qualitative result by using a cloud database server through a cloud storage module; and updating and displaying the acquired case information, the three-dimensional lung image, the lung three-dimensional image analysis result, the lung image, the lesion part characteristic point, the lung nodule screening model, the small-size image and the real-time data of the lung nodule qualitative result by using the display through the updating and displaying module.
Further, in the first step, the case information acquisition parameters further include a start time and an end time of the case information acquisition requirement instruction of the batch; when the server detects that the starting time is equal to the ending time, recording the current time, and setting the ending time as the current time.
Further, in the second step, the method for normalizing the three-dimensional lung image to be processed according to the sampling point includes:
(1) acquiring preset point coordinates, and converting the sampling model into an equation set containing preset parameters according to the preset point coordinates;
(2) assigning the preset parameters according to a preset rule, and when the current assignment enables the equation to be formed, taking the current assignment as the preset parameters to obtain a target sampling model;
(3) and carrying out nonlinear sampling on the iris image to be processed according to the target sampling model to obtain sampling points, and normalizing the image to be processed according to the sampling points.
Further, in step four, the method for deleting useless information contained in the image data includes:
and filling the original gray value on the route passed by the eraser with the gray value of the background, and converting the original DICOM data into an 8-bit BMP gray map through a harmonic mapping manner from qualitative to quantitative.
Further, in the fifth step, the method for segmenting the lung region in the lung three-dimensional image by the lung tissue segmentation module by using lung tissue segmentation includes:
(1) obtaining a preliminary lung tissue through coarse segmentation based on a threshold value, background voxel removal, contour completion and trachea tissue removal;
(2) on each slice, respectively counting the concavity and convexity of the outer contour edge point of the 2D connected domain to determine the region needing to be filled;
(3) and filling the region by combining with morphological operation to obtain a lung segmentation result.
Further, in the fifth step, the method for constructing the pulmonary nodule screening model by the screening model construction module by using the pulmonary nodule screening model construction program according to the labeled feature points includes:
determining a lesion part, performing region selection on the lesion part, performing feature extraction on the selected lesion region, screening out features with low redundancy and high correlation, and constructing a pulmonary nodule screening model.
Further, in step six, the method for performing lung nodule screening model training by using a model training program through a model training module includes:
acquiring data characteristics of training data; generating substitute data with the same feature dimension according to the data features; model training is performed based on the training data and the surrogate data.
Another object of the present invention is to provide a deep learning technique-based pulmonary nodule screening system using the deep learning technique-based pulmonary nodule screening method, where the deep learning technique-based pulmonary nodule screening system includes:
the information acquisition module is connected with the central control module and used for acquiring the case information through an information acquisition program to obtain the case information;
the lung image acquisition module is connected with the central control module and is used for acquiring the lung image of the patient through a CT (computed tomography) machine to obtain a three-dimensional lung image;
the image processing module is connected with the central control module and is used for carrying out normalization processing on the acquired lung three-dimensional image through an image processing program;
the central control module is connected with the information acquisition module, the lung image acquisition module, the image processing module, the image analysis module, the lung tissue segmentation module, the feature extraction module, the screening model construction module, the model training module, the image segmentation module, the lung nodule qualitative module, the cloud storage module and the updating display module and is used for controlling the normal operation of each module of the lung nodule screening system based on the deep learning technology through the central processing unit;
the image analysis module is connected with the central control module and is used for analyzing the processed lung three-dimensional image through an image analysis program;
the lung tissue segmentation module is connected with the central control module and used for segmenting the lung region in the lung three-dimensional image through lung tissue segmentation to obtain a lung image;
the characteristic extraction module is connected with the central control module and used for comparing the case result with the lung image through a characteristic extraction program to obtain and label the characteristic points of the pathological part;
the screening model construction module is connected with the central control module and used for constructing a pulmonary nodule screening model according to the labeled characteristic points through a pulmonary nodule screening model construction program;
the model training module is connected with the central control module and used for training the pulmonary nodule screening model through a model training program to obtain a trained pulmonary nodule screening model;
the image segmentation module is connected with the central control module and is used for segmenting the lung three-dimensional image through an image segmentation program to obtain a small-size image;
the pulmonary nodule qualitative module is connected with the central control module and used for screening the small-size image according to the trained pulmonary nodule screening model through a nodule qualitative program to obtain a pulmonary nodule qualitative result;
the cloud storage module is connected with the central control module and used for storing the acquired case information, the lung three-dimensional image analysis result, the lung image, the lesion part feature point, the lung nodule screening model, the small-size image and the lung nodule qualitative result through the cloud database server;
and the updating display module is connected with the central control module and is used for updating and displaying the acquired case information, the lung three-dimensional image analysis result, the lung image, the lesion part characteristic point, the lung nodule screening model, the small-size image and the real-time data of the lung nodule qualitative result through the display.
It is another object of the present invention to provide a computer program product stored on a computer readable medium, comprising a computer readable program for providing a user input interface to implement the method for deep learning technique based pulmonary nodule screening when executed on an electronic device.
It is another object of the present invention to provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method for pulmonary nodule screening based on deep learning techniques.
By combining all the technical schemes, the invention has the advantages and positive effects that: according to the lung nodule screening method based on the deep learning technology, the 2D space is used for performing edge completion through morphological operation according to the concave-convex property of the segmentation contour, so that nodules on the lung wall are prevented from being missed to be detected; the suspected nodule area is detected through lesion positioning based on local features, and is focused on a nodule and a similar area thereof, so that most normal lung tissues can be removed, and only the suspected nodular lesion area is reserved; the classification of various types of nodules can be realized through a focus classification method based on deep learning, the local three-dimensional information of the focus of the CT sequence image is fully utilized, whether the nodules are lung nodules or not is effectively distinguished, the types of the nodules are identified, and a doctor can be better assisted to improve the diagnosis accuracy; the screening efficiency of the pulmonary nodules is improved, and the false positive nodules are reduced.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments of the present application will be briefly described below, and it is obvious that the drawings described below are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained from the drawings without creative efforts.
Fig. 1 is a flowchart of a lung nodule screening method based on a deep learning technique according to an embodiment of the present invention.
FIG. 2 is a block diagram of a lung nodule screening system based on deep learning technology according to an embodiment of the present invention;
in the figure: 1. an information acquisition module; 2. a lung image acquisition module; 3. an image processing module; 4. a central control module; 5. an image analysis module; 6. a lung tissue segmentation module; 7. a feature extraction module; 8. a screening model construction module; 9. a model training module; 10. an image segmentation module; 11. a pulmonary nodule characterization module; 12. a cloud storage module; 13. and updating the display module.
Fig. 3 is a flowchart of a method for acquiring case information by an information acquisition module using an information acquisition program according to an embodiment of the present invention.
Fig. 4 is a flowchart of a method for normalizing an acquired three-dimensional lung image by an image processing module using an image processing program according to an embodiment of the present invention.
Fig. 5 is a flowchart of a method for analyzing a three-dimensional image of a lung after being processed by an image analysis module using an image analysis program according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In view of the problems in the prior art, the present invention provides a pulmonary nodule screening method and a screening system based on a deep learning technique, which are described in detail below with reference to the accompanying drawings.
As shown in fig. 1, a method for screening a pulmonary nodule based on a deep learning technique according to an embodiment of the present invention includes the following steps:
s101, acquiring case information by an information acquisition program through an information acquisition module to obtain the case information; acquiring a lung image of a patient by using a CT (computed tomography) machine through a lung image acquisition module to obtain a three-dimensional lung image;
s102, normalizing the acquired three-dimensional lung image by using an image processing program through an image processing module; the central control module utilizes a central processor to control the normal operation of each module of the pulmonary nodule screening system based on the deep learning technology;
s103, analyzing the processed lung three-dimensional image by using an image analysis program through an image analysis module; segmenting the lung region in the lung three-dimensional image by utilizing lung tissue segmentation through a lung tissue segmentation module to obtain a lung image;
s104, comparing the case result with the lung image by using a feature extraction program through a feature extraction module to obtain the feature points of the pathological part and marking the feature points;
s105, constructing a pulmonary nodule screening model by a screening model construction module by utilizing a pulmonary nodule screening model construction program according to the labeled characteristic points; training a pulmonary nodule screening model by using a model training program through a model training module to obtain a trained pulmonary nodule screening model;
s106, segmenting the lung three-dimensional image by using an image segmentation program through an image segmentation module to obtain a small-size image; screening the small-size image by using a nodule qualitative program through a pulmonary nodule qualitative module according to the trained pulmonary nodule screening model to obtain a pulmonary nodule qualitative result;
s107, storing the acquired case information, the three-dimensional lung image analysis result, the lung image, the lesion part feature point, the lung nodule screening model, the small-size image and the lung nodule qualitative result by using a cloud database server through a cloud storage module;
and S108, updating and displaying the acquired case information, the three-dimensional lung image, the lung three-dimensional image analysis result, the lung image, the lesion part feature point, the lung nodule screening model, the small-size image and the real-time data of the lung nodule qualitative result by using the display through the updating and displaying module.
In step S103 provided in the embodiment of the present invention, a method for segmenting a lung region in a three-dimensional lung image by a lung tissue segmentation module using lung tissue segmentation includes: obtaining a preliminary lung tissue through coarse segmentation based on a threshold value, background voxel removal, contour completion and trachea tissue removal; on each slice, respectively counting the concavity and convexity of the outer contour edge point of the 2D connected domain to determine the region needing to be filled; and filling the region by combining with morphological operation to obtain a lung segmentation result.
In step S105 provided in the embodiment of the present invention, a method for constructing a pulmonary nodule screening model by using a pulmonary nodule screening model construction program through a screening model construction module according to labeled feature points includes: determining a lesion part, performing region selection on the lesion part, performing feature extraction on the selected lesion region, screening out features with low redundancy and high correlation, and constructing a pulmonary nodule screening model.
In step S105 provided in the embodiment of the present invention, a method for performing lung nodule screening model training by using a model training program through a model training module includes: acquiring data characteristics of training data; generating substitute data with the same feature dimension according to the data features; model training is performed based on the training data and the surrogate data.
As shown in fig. 2, a lung nodule screening system based on deep learning technology provided by an embodiment of the present invention includes: the system comprises an information acquisition module 1, a lung image acquisition module 2, an image processing module 3, a central control module 4, an image analysis module 5, a lung tissue segmentation module 6, a feature extraction module 7, a screening model construction module 8, a model training module 9, an image segmentation module 10, a lung nodule qualitative module 11, a cloud storage module 12 and an update display module 13.
The information acquisition module 1 is connected with the central control module 4 and is used for acquiring case information through an information acquisition program to obtain the case information;
the lung image acquisition module 2 is connected with the central control module 4 and is used for acquiring the lung image of the patient through a CT (computed tomography) machine to obtain a three-dimensional lung image;
the image processing module 3 is connected with the central control module 4 and is used for carrying out normalization processing on the acquired lung three-dimensional image through an image processing program;
the central control module 4 is connected with the information acquisition module 1, the lung image acquisition module 2, the image processing module 3, the image analysis module 5, the lung tissue segmentation module 6, the feature extraction module 7, the screening model construction module 8, the model training module 9, the image segmentation module 10, the lung nodule qualitative module 11, the cloud storage module 12 and the update display module 13, and is used for controlling the normal operation of each module of the lung nodule screening system based on the deep learning technology through a central processing unit;
the image analysis module 5 is connected with the central control module 4 and is used for analyzing the processed lung three-dimensional image through an image analysis program;
the lung tissue segmentation module 6 is connected with the central control module 4 and used for segmenting the lung region in the lung three-dimensional image through lung tissue segmentation to obtain a lung image;
the characteristic extraction module 7 is connected with the central control module 4 and used for comparing the case result with the lung image through a characteristic extraction program to obtain and label the characteristic points of the pathological part;
the screening model building module 8 is connected with the central control module 4 and used for building a pulmonary nodule screening model according to the labeled characteristic points through a pulmonary nodule screening model building program;
the model training module 9 is connected with the central control module 4 and used for training the pulmonary nodule screening model through a model training program to obtain a trained pulmonary nodule screening model;
the image segmentation module 10 is connected with the central control module 4 and is used for segmenting the lung three-dimensional image through an image segmentation program to obtain a small-size image;
the lung nodule qualitative module 11 is connected with the central control module 4 and is used for screening the small-size image according to the trained lung nodule screening model through a nodule qualitative program to obtain a lung nodule qualitative result;
the cloud storage module 12 is connected with the central control module 4 and used for storing the acquired case information, the lung three-dimensional image analysis result, the lung image, the lesion part feature point, the lung nodule screening model, the small-size image and the lung nodule qualitative result through a cloud database server;
and the updating display module 13 is connected with the central control module 4 and is used for updating and displaying the acquired case information, the three-dimensional lung image, the lung three-dimensional image analysis result, the lung image, the lesion part feature point, the lung nodule screening model, the small-size image and the real-time data of the lung nodule qualitative result through a display.
The invention is further described with reference to specific examples.
Example 1
Fig. 1 shows a method for screening pulmonary nodules based on a deep learning technique according to an embodiment of the present invention, and fig. 3 shows a preferred embodiment of the method for acquiring case information by an information acquisition module using an information acquisition program according to an embodiment of the present invention, where the method includes:
s201, setting case information acquisition parameters at a terminal, and sending an information acquisition request carrying the case information acquisition parameters to a server through an information acquisition module;
s202, a server receives an information acquisition request sent by a terminal, acquires case information acquisition parameters in the information acquisition request, and acquires a target case information set in a database based on the case information acquisition parameters;
s203, the server sets case information feedback parameters and sends a target case information set and the case information feedback parameters to the terminal;
and S204, the terminal receives and stores the target case information set and the case information feedback parameters, sets the case information acquisition parameters based on the case information feedback parameters through the information acquisition program, and sends a case information acquisition request carrying the case information acquisition parameters to the server to obtain the case information.
The case information acquisition parameters provided by the embodiment of the invention also comprise the starting time and the ending time of the case information acquisition requirement instruction of the batch; when the server detects that the starting time is equal to the ending time, recording the current time, and setting the ending time as the current time.
Example 2
Fig. 1 shows a method for screening lung nodules based on a deep learning technique according to an embodiment of the present invention, and as a preferred embodiment, fig. 4 shows a method for normalizing acquired three-dimensional images of lungs by an image processing module using an image processing program according to an embodiment of the present invention, which includes:
s301, acquiring a lung image of a patient by using a CT (computed tomography) machine through a lung image acquisition module to obtain a three-dimensional lung image;
s302, acquiring a lung three-dimensional image to be processed through an image processing module, and determining a state type of the image to be processed;
s303, searching a corresponding sampling model according to the state type by using an image processing program;
s304, carrying out nonlinear sampling on the image to be processed according to the sampling model to obtain a sampling point, and carrying out normalization processing on the lung three-dimensional image to be processed according to the sampling point.
The method for normalizing the lung three-dimensional image to be processed according to the sampling point, provided by the embodiment of the invention, comprises the following steps:
(1) acquiring preset point coordinates, and converting the sampling model into an equation set containing preset parameters according to the preset point coordinates;
(2) assigning the preset parameters according to a preset rule, and when the current assignment enables the equation to be formed, taking the current assignment as the preset parameters to obtain a target sampling model;
(3) and carrying out nonlinear sampling on the iris image to be processed according to the target sampling model to obtain sampling points, and normalizing the image to be processed according to the sampling points.
Example 3
Fig. 1 shows a method for screening lung nodules based on a deep learning technique according to an embodiment of the present invention, and fig. 5 shows a preferred embodiment of the method for post-treatment three-dimensional image analysis of lungs by an image analysis module using an image analysis program according to an embodiment of the present invention, where the method includes:
s401, image preprocessing is carried out by using an image analysis program, the image preprocessing is completed by image filtering processing, target sub-image acquisition and histogram equalization processing, and useless information contained in image data is deleted to obtain image information of a region to be treated;
s402, image segmentation and automatic quantitative analysis are carried out to obtain external contour features of the target, the number of pixels contained in the longest distance and the widest distance of a target area is respectively obtained on the basis of image segmentation, and the pixels are converted into length and width values of an entity according to a scale coefficient to complete three-dimensional visual angle conversion;
and S403, displaying tissues of different parts according to the interactive region selection, realizing the analysis of the processed lung three-dimensional image, and generating a lung three-dimensional image analysis report.
The method for deleting the useless information contained in the image data provided by the embodiment of the invention comprises the following steps: and filling the original gray value on the route passed by the eraser with the gray value of the background, and converting the original DICOM data into an 8-bit BMP gray map through a harmonic mapping manner from qualitative to quantitative.
In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When used in whole or in part, can be implemented in a computer program product that includes one or more computer instructions. When loaded or executed on a computer, cause the flow or functions according to embodiments of the invention to occur, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, the computer instructions may be transmitted from one website site, computer, server, or data center to another website site, computer, server, or data center via wire (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL), or wireless (e.g., infrared, wireless, microwave, etc.)). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device, such as a server, a data center, etc., that includes one or more of the available media. The usable medium may be a magnetic medium (e.g., floppy Disk, hard Disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., Solid State Disk (SSD)), among others.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention, and the scope of the present invention is not limited thereto, and any modification, equivalent replacement, and improvement made by those skilled in the art within the technical scope of the present invention disclosed herein, which is within the spirit and principle of the present invention, should be covered by the present invention.
Claims (10)
1. A pulmonary nodule screening method based on a deep learning technology is characterized by comprising the following steps:
setting a case information acquisition parameter at a terminal, and sending an information acquisition request carrying the case information acquisition parameter to a server through an information acquisition module; the server receives an information acquisition request sent by the terminal, acquires case information acquisition parameters in the information acquisition request, and acquires a target case information set in the database based on the case information acquisition parameters; the server sets case information feedback parameters and sends a target case information set and the case information feedback parameters to the terminal; the terminal receives and stores a target case information set and case information feedback parameters, sets case information acquisition parameters based on the case information feedback parameters through an information acquisition program, and sends a case information acquisition request carrying the case information acquisition parameters to a server to obtain case information;
acquiring a lung image of the patient by using a CT machine through a lung image acquisition module to obtain a three-dimensional lung image; acquiring a lung three-dimensional image to be processed through an image processing module, and determining a state type of the image to be processed; searching a corresponding sampling model according to the state type by using an image processing program; carrying out nonlinear sampling on the image to be processed according to the sampling model to obtain a sampling point, and carrying out normalization processing on the lung three-dimensional image to be processed according to the sampling point;
step three, the server terminal adopts a template matching algorithm based on shape characteristics and a detection algorithm based on SIFT characteristics for detection, calls an image processing operator packaged in an OpenCV visual class library through an image analysis module, programs corresponding C + + codes, and realizes the function of the algorithm by utilizing each control button in an application software interface;
step four, image preprocessing is carried out by utilizing an image analysis program, image filtering processing, target sub-image acquisition and histogram equalization processing are completed by the image preprocessing, useless information contained in the image data is deleted, and image information of the area to be treated is obtained; image segmentation and automatic quantitative analysis are carried out to obtain the external contour characteristics of the target, on the basis of image segmentation, the number of pixels contained in the longest and widest distances of a target area is respectively obtained, and the pixels are converted into the length and width values of an entity according to a scale coefficient to complete three-dimensional visual angle conversion; displaying tissues of different parts according to interactive region selection, realizing analysis of the processed three-dimensional lung image, and generating a lung three-dimensional image analysis report;
fifthly, segmenting the lung region in the lung three-dimensional image by utilizing lung tissue segmentation through a lung tissue segmentation module to obtain a lung image; comparing the case result with the lung image by using a feature extraction program through a feature extraction module to obtain and label the feature points of the pathological part; constructing a pulmonary nodule screening model according to the labeled characteristic points by using a screening model construction module and a pulmonary nodule screening model construction program;
step six, training the pulmonary nodule screening model by using a model training program through a model training module to obtain a trained pulmonary nodule screening model; the image segmentation module is used for segmenting the lung three-dimensional image by using an image segmentation program to obtain a small-size image; screening the small-size image by using a nodule qualitative program through a pulmonary nodule qualitative module according to the trained pulmonary nodule screening model to obtain a pulmonary nodule qualitative result;
step seven, storing the acquired case information, the lung three-dimensional image analysis result, the lung image, the lesion part feature point, the lung nodule screening model, the small-size image and the lung nodule qualitative result by using a cloud database server through a cloud storage module; and updating and displaying the acquired case information, the three-dimensional lung image, the lung three-dimensional image analysis result, the lung image, the lesion part characteristic point, the lung nodule screening model, the small-size image and the real-time data of the lung nodule qualitative result by using the display through the updating and displaying module.
2. The method for screening pulmonary nodules based on deep learning technology as claimed in claim 1, wherein in step one, the case information acquisition parameters further include a start time and an end time of the case information acquisition requirement indication of the batch; when the server detects that the starting time is equal to the ending time, recording the current time, and setting the ending time as the current time.
3. The method for screening pulmonary nodules based on deep learning technology as claimed in claim 1, wherein in step two, the method for normalizing the three-dimensional image of the lung to be processed according to the sampling points includes:
(1) acquiring preset point coordinates, and converting the sampling model into an equation set containing preset parameters according to the preset point coordinates;
(2) assigning the preset parameters according to a preset rule, and when the current assignment enables the equation to be formed, taking the current assignment as the preset parameters to obtain a target sampling model;
(3) and carrying out nonlinear sampling on the iris image to be processed according to the target sampling model to obtain sampling points, and normalizing the image to be processed according to the sampling points.
4. The method for screening pulmonary nodules based on deep learning technology as claimed in claim 1, wherein in step four, the method for removing useless information contained in the image data comprises:
and filling the original gray value on the route passed by the eraser with the gray value of the background, and converting the original DICOM data into an 8-bit BMP gray map through a harmonic mapping manner from qualitative to quantitative.
5. The method for screening pulmonary nodules based on deep learning technology as claimed in claim 1, wherein in step five, the method for segmenting lung regions in the lung three-dimensional image by using lung tissue segmentation module includes:
(1) obtaining a preliminary lung tissue through coarse segmentation based on a threshold value, background voxel removal, contour completion and trachea tissue removal;
(2) on each slice, respectively counting the concavity and convexity of the outer contour edge point of the 2D connected domain to determine the region needing to be filled;
(3) and filling the region by combining with morphological operation to obtain a lung segmentation result.
6. The method for screening lung nodules based on deep learning technology as claimed in claim 1, wherein in step five, the method for screening lung nodules by using the screening model building module to build the lung nodule screening model according to the labeled feature points by using the lung nodule screening model building program includes:
determining a lesion part, performing region selection on the lesion part, performing feature extraction on the selected lesion region, screening out features with low redundancy and high correlation, and constructing a pulmonary nodule screening model.
7. The method for screening pulmonary nodules based on deep learning technology as claimed in claim 1, wherein in step six, the method for training pulmonary nodule screening model by model training module using model training program includes:
acquiring data characteristics of training data; generating substitute data with the same feature dimension according to the data features; model training is performed based on the training data and the surrogate data.
8. A deep learning technique-based pulmonary nodule screening system using the deep learning technique-based pulmonary nodule screening method according to any one of claims 1 to 7, wherein the deep learning technique-based pulmonary nodule screening system comprises:
the information acquisition module is connected with the central control module and used for acquiring the case information through an information acquisition program to obtain the case information;
the lung image acquisition module is connected with the central control module and is used for acquiring the lung image of the patient through a CT (computed tomography) machine to obtain a three-dimensional lung image;
the image processing module is connected with the central control module and is used for carrying out normalization processing on the acquired lung three-dimensional image through an image processing program;
the central control module is connected with the information acquisition module, the lung image acquisition module, the image processing module, the image analysis module, the lung tissue segmentation module, the feature extraction module, the screening model construction module, the model training module, the image segmentation module, the lung nodule qualitative module, the cloud storage module and the updating display module and is used for controlling the normal operation of each module of the lung nodule screening system based on the deep learning technology through the central processing unit;
the image analysis module is connected with the central control module and is used for analyzing the processed lung three-dimensional image through an image analysis program;
the lung tissue segmentation module is connected with the central control module and used for segmenting the lung region in the lung three-dimensional image through lung tissue segmentation to obtain a lung image;
the characteristic extraction module is connected with the central control module and used for comparing the case result with the lung image through a characteristic extraction program to obtain and label the characteristic points of the pathological part;
the screening model construction module is connected with the central control module and used for constructing a pulmonary nodule screening model according to the labeled characteristic points through a pulmonary nodule screening model construction program;
the model training module is connected with the central control module and used for training the pulmonary nodule screening model through a model training program to obtain a trained pulmonary nodule screening model;
the image segmentation module is connected with the central control module and is used for segmenting the lung three-dimensional image through an image segmentation program to obtain a small-size image;
the pulmonary nodule qualitative module is connected with the central control module and used for screening the small-size image according to the trained pulmonary nodule screening model through a nodule qualitative program to obtain a pulmonary nodule qualitative result;
the cloud storage module is connected with the central control module and used for storing the acquired case information, the lung three-dimensional image analysis result, the lung image, the lesion part feature point, the lung nodule screening model, the small-size image and the lung nodule qualitative result through the cloud database server;
and the updating display module is connected with the central control module and is used for updating and displaying the acquired case information, the lung three-dimensional image analysis result, the lung image, the lesion part characteristic point, the lung nodule screening model, the small-size image and the real-time data of the lung nodule qualitative result through the display.
9. A computer program product stored on a computer readable medium, comprising a computer readable program for providing a user input interface for implementing a method of deep learning technique based lung nodule screening as claimed in any one of claims 1 to 7 when executed on an electronic device.
10. A computer readable storage medium storing instructions which, when executed on a computer, cause the computer to perform a method of pulmonary nodule screening based on deep learning techniques as claimed in any one of claims 1 to 7.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110368194.0A CN113160153A (en) | 2021-04-06 | 2021-04-06 | Lung nodule screening method and system based on deep learning technology |
| ZA2022/01219A ZA202201219B (en) | 2021-04-06 | 2022-01-26 | Lung nodule screening method and screening system based on deep learning technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110368194.0A CN113160153A (en) | 2021-04-06 | 2021-04-06 | Lung nodule screening method and system based on deep learning technology |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113160153A true CN113160153A (en) | 2021-07-23 |
Family
ID=76888606
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110368194.0A Pending CN113160153A (en) | 2021-04-06 | 2021-04-06 | Lung nodule screening method and system based on deep learning technology |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN113160153A (en) |
| ZA (1) | ZA202201219B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113506296A (en) * | 2021-09-10 | 2021-10-15 | 之江实验室 | Slow obstructive pulmonary disease diagnosis device based on priori knowledge CT (computed tomography) subregion image omics |
| CN113610829A (en) * | 2021-08-17 | 2021-11-05 | 大连大学附属中山医院 | System for screening pulmonary nodules based on artificial intelligence detection |
| CN113903452A (en) * | 2021-09-18 | 2022-01-07 | 宁波大学医学院附属医院 | Auxiliary diagnosis system for pulmonary nodules of doctors |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106780460A (en) * | 2016-12-13 | 2017-05-31 | 杭州健培科技有限公司 | A kind of Lung neoplasm automatic checkout system for chest CT image |
| CN108288052A (en) * | 2018-03-01 | 2018-07-17 | 武汉轻工大学 | Iris image method for normalizing, device and computer readable storage medium |
| CN110211143A (en) * | 2019-06-10 | 2019-09-06 | 张家港江苏科技大学产业技术研究院 | A kind of medical image analysis method based on computer vision |
| CN111369530A (en) * | 2020-03-04 | 2020-07-03 | 浙江明峰智能医疗科技有限公司 | CT image pulmonary nodule rapid screening method based on deep learning |
| CN111882538A (en) * | 2020-07-28 | 2020-11-03 | 杭州市红十字会医院 | Processing system and information processing method for distinguishing pulmonary tuberculosis information and tumor information |
-
2021
- 2021-04-06 CN CN202110368194.0A patent/CN113160153A/en active Pending
-
2022
- 2022-01-26 ZA ZA2022/01219A patent/ZA202201219B/en unknown
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106780460A (en) * | 2016-12-13 | 2017-05-31 | 杭州健培科技有限公司 | A kind of Lung neoplasm automatic checkout system for chest CT image |
| CN108288052A (en) * | 2018-03-01 | 2018-07-17 | 武汉轻工大学 | Iris image method for normalizing, device and computer readable storage medium |
| CN110211143A (en) * | 2019-06-10 | 2019-09-06 | 张家港江苏科技大学产业技术研究院 | A kind of medical image analysis method based on computer vision |
| CN111369530A (en) * | 2020-03-04 | 2020-07-03 | 浙江明峰智能医疗科技有限公司 | CT image pulmonary nodule rapid screening method based on deep learning |
| CN111882538A (en) * | 2020-07-28 | 2020-11-03 | 杭州市红十字会医院 | Processing system and information processing method for distinguishing pulmonary tuberculosis information and tumor information |
Non-Patent Citations (1)
| Title |
|---|
| 栗科峰: "《人脸图像处理与识别技术》", 30 September 2018, pages: 30 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113610829A (en) * | 2021-08-17 | 2021-11-05 | 大连大学附属中山医院 | System for screening pulmonary nodules based on artificial intelligence detection |
| CN113506296A (en) * | 2021-09-10 | 2021-10-15 | 之江实验室 | Slow obstructive pulmonary disease diagnosis device based on priori knowledge CT (computed tomography) subregion image omics |
| CN113903452A (en) * | 2021-09-18 | 2022-01-07 | 宁波大学医学院附属医院 | Auxiliary diagnosis system for pulmonary nodules of doctors |
| CN113903452B (en) * | 2021-09-18 | 2022-12-16 | 宁波大学医学院附属医院 | Auxiliary diagnosis system for pulmonary nodules of doctors |
Also Published As
| Publication number | Publication date |
|---|---|
| ZA202201219B (en) | 2022-03-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111127466B (en) | Medical image detection method, device, equipment and storage medium | |
| CN113160153A (en) | Lung nodule screening method and system based on deep learning technology | |
| CN108133476B (en) | Method and system for automatically detecting pulmonary nodules | |
| JP2002523123A (en) | Method and system for lesion segmentation and classification | |
| JP2008520345A (en) | Method and system for detecting and classifying lesions in ultrasound images | |
| US20210327583A1 (en) | Determination of a growth rate of an object in 3d data sets using deep learning | |
| CN114092450A (en) | Real-time image segmentation method, system and device based on gastroscopy video | |
| CN111353978B (en) | Method and device for identifying heart anatomy structure | |
| CN113096109A (en) | Lung medical image analysis method, device and system | |
| US20080075345A1 (en) | Method and System For Lymph Node Segmentation In Computed Tomography Images | |
| CN115601811A (en) | Facial acne detection method and device | |
| CN110648318A (en) | Auxiliary analysis method and device for skin diseases, electronic equipment and storage medium | |
| WO2020168647A1 (en) | Image recognition method and related device | |
| Tian et al. | Radiomics and Its Clinical Application: Artificial Intelligence and Medical Big Data | |
| CN113269721A (en) | Model training method and device, electronic equipment and storage medium | |
| CN111612749B (en) | Focus detection method and device based on lung image | |
| CN116309647B (en) | Method for constructing craniocerebral lesion image segmentation model, image segmentation method and device | |
| CN114037719B (en) | Bone region determination method and device, electronic equipment and storage medium | |
| CN112634240B (en) | Thyroid ultrasound image interference interface automatic identification and removal method based on threshold segmentation | |
| CN115147359A (en) | Lung lobe segmentation network model training method and device, electronic equipment and storage medium | |
| CN114663443A (en) | 12-lead paper electrocardiogram digitization method and device | |
| CN114187281A (en) | Image processing method and device, electronic equipment and storage medium | |
| CN113658198A (en) | Interactive emphysema focus segmentation method, device, storage medium and equipment | |
| CN113222009A (en) | Method for estimating nodule aspect ratio based on thyroid ultrasound video | |
| CN117058292B (en) | Tone scale map rendering system based on digital pathological image |
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 |