CN111724347A - Chest radiography image anomaly detection method and system based on deep learning - Google Patents
Chest radiography image anomaly detection method and system based on deep learning Download PDFInfo
- Publication number
- CN111724347A CN111724347A CN202010443658.5A CN202010443658A CN111724347A CN 111724347 A CN111724347 A CN 111724347A CN 202010443658 A CN202010443658 A CN 202010443658A CN 111724347 A CN111724347 A CN 111724347A
- Authority
- CN
- China
- Prior art keywords
- chest
- image
- chest radiography
- abnormal
- images
- 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.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/60—Protecting data
- G06F21/62—Protecting access to data via a platform, e.g. using keys or access control rules
- G06F21/6218—Protecting access to data via a platform, e.g. using keys or access control rules to a system of files or objects, e.g. local or distributed file system or database
- G06F21/6245—Protecting personal data, e.g. for financial or medical purposes
- G06F21/6254—Protecting personal data, e.g. for financial or medical purposes by anonymising data, e.g. decorrelating personal data from the owner's identification
-
- 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
- G06T2207/10116—X-ray image
-
- 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
- G06T2207/30204—Marker
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Data Mining & Analysis (AREA)
- Bioethics (AREA)
- General Engineering & Computer Science (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Quality & Reliability (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Computation (AREA)
- Radiology & Medical Imaging (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Databases & Information Systems (AREA)
- Computer Hardware Design (AREA)
- Computer Security & Cryptography (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a chest radiography image abnormity detection method and system based on deep learning, wherein the method comprises the following steps: constructing a basic identification algorithm; obtaining a chest radiography image, and learning the chest radiography image according to a basic identification algorithm to generate a classification model; classifying the chest radiography images to be detected according to the classification model to obtain abnormal chest radiography images and normal chest radiography images; and marking the abnormal chest radiography image to finish the detection. The chest radiography images are deeply learned by constructing a basic identification algorithm so as to generate a classification model, the chest radiography images to be detected can be directly classified to obtain normal chest radiography images and abnormal chest radiography images, then the abnormal chest radiography images are sent to a doctor, and the normal chest radiography images are not sent to the doctor, so that the working content of the doctor is reduced, and the working efficiency of the doctor is improved; in addition, by labeling the abnormal chest radiography image, a doctor can intuitively observe the position and the reason of the abnormality of the chest radiography image.
Description
Technical Field
The invention relates to the technical field of chest radiography image detection, in particular to a chest radiography image abnormity detection method and system based on deep learning.
Background
With the development and popularization of radiography technology, more and more medical institutions introduce X-ray equipment, particularly DR equipment, and are widely applied to various business scenes of various medical institutions, wherein the most common and most numerous chest radiography images are available, and the patient information can be identified by analyzing the chest radiography images.
However, the direct transmission of the chest image taken by the patient to the doctor for analysis has the following disadvantages: (1) after the doctor takes the chest image, the doctor can judge whether the chest image is abnormal or not by depending on the professional knowledge and medical experience of the doctor, so that the doctor cannot know where the abnormality exists immediately if the experience is not enough, but the cost for training a high-level doctor is high; (2) in general, normal chest radiographs are abundant, and the normal chest radiographs do not need any analysis by a doctor actually, so that the time of the doctor is wasted and the work efficiency of the doctor is affected when the normal chest radiographs are directly sent to the doctor.
Disclosure of Invention
The invention aims to provide a chest radiography image abnormity detection method and system based on deep learning, and solves the problems that the position of the chest radiography image abnormity cannot be judged and the working efficiency of doctors is low in the prior art.
The invention is realized by the following technical scheme:
a chest radiography image abnormity detection method based on deep learning specifically comprises the following steps:
step S1, constructing a basic identification algorithm for deep learning;
step S2, obtaining a chest radiography image, and performing deep learning on the chest radiography image according to the basic identification algorithm to generate a classification model;
step S3, classifying the chest radiography image to be detected according to the classification model to obtain an abnormal chest radiography image and a normal chest radiography image;
and step S4, labeling the abnormal chest radiography image to finish detection.
As a further alternative of the chest image anomaly detection method based on deep learning, the method is based on a chest image feature database, the chest image feature database comprises chest image basic identification features, chest image exclusion identification features and chest image anomaly identification features, and the basic identification algorithm is constructed according to the chest image basic identification features, the chest image exclusion identification features and the chest image anomaly identification features.
As a further alternative of the chest image abnormality detection method based on deep learning, the step S2 includes the steps of:
step S21, acquiring a normal chest image and an abnormal chest image, and desensitizing the acquired chest images;
and step S22, learning the chest radiography image after desensitization treatment according to a basic identification algorithm, thereby generating a classification model.
As a further alternative of the chest image abnormality detection method based on deep learning, the step S3 includes the steps of:
step S31, inputting the chest picture image to be detected into the classification model;
step S32, the classification model carries out feature extraction on the input chest picture image to be detected;
and step S33, the classification model judges the extracted features and judges whether the features are abnormal identification features of the chest radiography images, if so, the chest radiography images are classified as abnormal chest radiography images, and if not, the chest radiography images are classified as normal chest radiography images.
As a further alternative of the chest image abnormality detection method based on deep learning, the step S4 includes the steps of:
step S41, segmenting an image of the position of the abnormal identification feature in the abnormal chest radiography image by adopting a threshold value method;
step S42, extracting the features of the image at the position of the abnormal identification features, and extracting the abnormal identification features;
and step S43, labeling the abnormal identification features to finish detection.
A chest radiography image abnormity detection system based on deep learning utilizes any one of the detection methods.
The invention has the beneficial effects that:
by using the method and the system, the chest radiography images are deeply learned by constructing a basic identification algorithm, so that a classification model is generated, and the chest radiography images to be detected can be directly judged to be normal or abnormal, so that the normal chest radiography images and the abnormal chest radiography images are classified, the abnormal chest radiography images are sent to a doctor, and the normal chest radiography images are not sent to the doctor, so that the working content of the doctor is reduced, the working efficiency of the doctor is improved, and the problem of low working efficiency of the doctor in the prior art is solved; in addition, by labeling the abnormal chest radiography images, doctors can intuitively observe where the abnormal positions of the chest radiography images are, so that the reasons of the abnormal chest radiography images can be quickly determined, and the problem that the abnormal positions of the chest radiography images cannot be judged if the experience of the doctors is insufficient in the prior art is solved.
Drawings
Fig. 1 is a schematic flow chart of a real-time detection method of two-hand gestures according to the present invention.
Detailed Description
The invention will be described in detail with reference to the drawings and specific embodiments, which are illustrative of the invention and are not to be construed as limiting the invention.
As shown in fig. 1, a chest radiography image abnormality detection method based on deep learning specifically includes the following steps:
step S1, constructing a basic identification algorithm for deep learning;
step S2, obtaining a chest radiography image, and performing deep learning on the chest radiography image according to the basic identification algorithm to generate a classification model;
step S3, classifying the chest radiography image to be detected according to the classification model to obtain an abnormal chest radiography image and a normal chest radiography image;
and step S4, labeling the abnormal chest radiography image to finish detection.
In the embodiment, the chest radiography images are deeply learned by constructing a basic identification algorithm so as to generate a classification model, and the normal or abnormal judgment can be directly carried out on the chest radiography images to be detected, so that the normal chest radiography images and the abnormal chest radiography images are classified, then the abnormal chest radiography images are sent to a doctor, and the normal chest radiography images are not sent to the doctor, so that the working content of the doctor is reduced, the working efficiency of the doctor is improved, and the problem of low working efficiency of the doctor in the prior art is solved; in addition, by labeling the abnormal chest radiography images, doctors can intuitively observe where the abnormal positions of the chest radiography images are, so that the reasons of the abnormal chest radiography images can be quickly determined, and the problem that the abnormal positions of the chest radiography images cannot be judged if the experience of the doctors is insufficient in the prior art is solved.
It should be noted that the chest image obtained in step S2 may be divided into training data used for deep learning of the basic recognition algorithm in step S2 and test data used for testing the classification performance of the classification model as an input of the classification model in step S3.
Preferably, the method is based on a chest image feature database, the chest image feature database comprises chest image basic identification features, chest image exclusion identification features and chest image abnormity identification features, and the basic identification algorithm is constructed according to the chest image basic identification features, the chest image exclusion identification features and the chest image abnormity identification features.
In the embodiment, a basic identification algorithm is formed by acquiring a chest image basic identification feature, a chest image exclusion identification feature and a chest image abnormal identification feature which are pre-stored in a chest image feature database, and eliminating interference data on the basis of the basic identification feature and identifying the abnormal identification feature, so that a chest image basic identification algorithm for deep learning can be constructed;
it should be noted that, according to the difference of medical knowledge, gender and age, the feature points of different chest X-ray films are pre-stored in the chest image feature database as basic identification features, so that the sample diversity for constructing a basic identification algorithm can be increased; according to the experience of doctors and the condition of hospital equipment, artifact characteristics appearing in images caused by the damage of patient clothes, equipment components, inaccurate calibration and the like are pre-stored in a chest image characteristic database to be taken as exclusion identification characteristics, and the identification capability of a basic identification algorithm can be improved by excluding the identification characteristics; according to the image medical expert team, pre-storing feature points marked with abnormal conditions and condition descriptions of abnormal films in a chest image feature database as abnormal identification features; in addition, the content of the chest image feature database can be modified or increased according to the characteristics of the image, so that the effect of calibrating the basic identification algorithm is achieved.
Preferably, the step S2 includes the steps of:
step S21, acquiring a normal chest image and an abnormal chest image, and desensitizing the acquired chest images;
and step S22, learning the chest radiography image after desensitization treatment according to a basic identification algorithm, thereby generating a classification model.
In the embodiment, firstly, the chest radiograph images of a hospital are sorted, the chest radiograph images are classified according to the primary diagnosis information of a doctor and divided into normal chest radiograph images and abnormal chest radiograph images, then the normal chest radiograph images and the abnormal chest radiograph images are desensitized, redundant information on the chest radiograph images is removed, the redundant information includes but is not limited to personal information and image numbers of patients, and then the chest radiograph images after desensitization are deeply learned through a constructed basic identification algorithm, so that a classification network capable of judging whether the chest radiograph images are abnormal or not can be obtained;
it should be noted that, by performing desensitization processing on the chest radiography image, redundant information on the chest radiography image is removed, and the judgment capability of the classification network can be improved.
Preferably, the step S3 includes the steps of:
step S31, inputting the chest picture image to be detected into the classification model;
step S32, the classification model carries out feature extraction on the input chest picture image to be detected;
and step S33, the classification model judges the extracted features and judges whether the features are abnormal identification features of the chest radiography images, if so, the chest radiography images are classified as abnormal chest radiography images, and if not, the chest radiography images are classified as normal chest radiography images.
In the embodiment, the classification model is used for extracting the features of the chest image to be detected firstly and then judging the extracted features, so that a normal chest image and an abnormal chest image are classified, the classification effect of the classification network is more remarkable, and the classification network can be judged more flexibly through the chest image features because the basic identification algorithm is constructed based on the chest image basic identification features, the chest image exclusion identification features and the chest image abnormal identification features, and the classification model is generated based on the basic identification algorithm.
Preferably, the step S4 includes the steps of:
step S41, segmenting an image of the position of the abnormal identification feature in the abnormal chest radiography image by adopting a threshold value method;
step S42, extracting the features of the image at the position of the abnormal identification features, and extracting the abnormal identification features;
and step S43, labeling the abnormal identification features to finish detection.
In this embodiment, since a plurality of abnormal identification features may exist in one abnormal chest radiograph image, an image at the position where each abnormal identification feature is located needs to be segmented by a threshold method, and after the segmentation is completed, feature extraction is performed on the image at the position where each abnormal identification feature is located, so as to extract the abnormal identification feature of the image, and since the abnormal identification feature includes the abnormal condition and the abnormal reason of the abnormal movie, a doctor can intuitively observe which position has an abnormality and can know the reason of the abnormality only by labeling the extracted abnormal identification feature;
it should be noted that, the labeling in labeling the abnormality identification features includes, but is not limited to, labeling the abnormality identification features through colors, different abnormality identification features can be labeled through different colors, a color explanatory diagram is preset, and a doctor can pointedly find whether corresponding colors are labeled on the abnormal chest radiograph image through the color explanatory diagram, so that the radiograph reading speed of the doctor is greatly improved, and the radiograph reading time of the doctor is shortened.
A chest radiography image abnormity detection system based on deep learning utilizes any one of the detection methods.
In the embodiment, the chest radiography images are deeply learned by constructing a basic identification algorithm so as to generate a classification model, and the normal or abnormal judgment can be directly carried out on the chest radiography images to be detected, so that the normal chest radiography images and the abnormal chest radiography images are classified, then the abnormal chest radiography images are sent to a doctor, and the normal chest radiography images are not sent to the doctor, so that the working content of the doctor is reduced, the working efficiency of the doctor is improved, and the problem of low working efficiency of the doctor in the prior art is solved; in addition, by labeling the abnormal chest radiography images, doctors can intuitively observe where the abnormal positions of the chest radiography images are, so that the reasons of the abnormal chest radiography images can be quickly determined, and the problem that the abnormal positions of the chest radiography images cannot be judged if the experience of the doctors is insufficient in the prior art is solved.
It should be noted that the obtained chest radiography images are divided into training data and test data, the training data is used for deep learning of the basic identification algorithm, and the test data is used as the input of the classification model to test the classification performance of the classification model.
The technical solutions provided by the embodiments of the present invention are described in detail above, and the principles and embodiments of the present invention are explained herein by using specific examples, and the descriptions of the embodiments are only used to help understanding the principles of the embodiments of the present invention; meanwhile, for a person skilled in the art, according to the embodiments of the present invention, there may be variations in the specific implementation manners and application ranges, and in summary, the content of the present description should not be construed as a limitation to the present invention.
Claims (6)
1. A chest radiography image abnormity detection method based on deep learning is characterized in that: the method specifically comprises the following steps:
step S1, constructing a basic identification algorithm for deep learning;
step S2, obtaining a chest radiography image, and performing deep learning on the chest radiography image according to the basic identification algorithm to generate a classification model;
step S3, classifying the chest radiography image to be detected according to the classification model to obtain an abnormal chest radiography image and a normal chest radiography image;
and step S4, labeling the abnormal chest radiography image to finish detection.
2. The method for detecting the abnormality of the chest radiography image based on the deep learning of claim 1, wherein: the method is based on a chest image feature database, the chest image feature database comprises chest image basic identification features, chest image exclusion identification features and chest image abnormity identification features, and the basic identification algorithm is constructed according to the chest image basic identification features, the chest image exclusion identification features and the chest image abnormity identification features.
3. The method for detecting the abnormality of the chest radiography image based on the deep learning of claim 2, wherein: the step S2 includes the steps of:
step S21, acquiring a normal chest image and an abnormal chest image, and desensitizing the acquired chest images;
and step S22, learning the chest radiography image after desensitization treatment according to a basic identification algorithm, thereby generating a classification model.
4. The method for detecting abnormality of chest image based on deep learning as claimed in claim 1 or 3, wherein: the step S3 includes the steps of:
step S31, inputting the chest picture image to be detected into the classification model;
step S32, the classification model carries out feature extraction on the input detected chest radiography image;
and step S33, the classification model judges the extracted features and judges whether the features are abnormal identification features of the chest radiography images, if so, the chest radiography images are classified as abnormal chest radiography images, and if not, the chest radiography images are classified as normal chest radiography images.
5. The method for detecting the abnormality of the chest radiography image based on the deep learning of claim 4, wherein: the step S4 includes the steps of:
step S41, segmenting an image of the position of the abnormal identification feature in the abnormal chest radiography image by adopting a threshold value method;
step S42, extracting the features of the image at the position of the abnormal identification features, and extracting the abnormal identification features;
and step S43, labeling the abnormal identification features to finish detection.
6. The utility model provides a chest radiography image anomaly detection system based on deep learning which characterized in that: the system uses the detection method of any one of claims 1 to 5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010443658.5A CN111724347A (en) | 2020-05-22 | 2020-05-22 | Chest radiography image anomaly detection method and system based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010443658.5A CN111724347A (en) | 2020-05-22 | 2020-05-22 | Chest radiography image anomaly detection method and system based on deep learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111724347A true CN111724347A (en) | 2020-09-29 |
Family
ID=72564910
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010443658.5A Withdrawn CN111724347A (en) | 2020-05-22 | 2020-05-22 | Chest radiography image anomaly detection method and system based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111724347A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112308853A (en) * | 2020-10-20 | 2021-02-02 | 平安科技(深圳)有限公司 | Electronic equipment, medical image index generation method and device and storage medium |
-
2020
- 2020-05-22 CN CN202010443658.5A patent/CN111724347A/en not_active Withdrawn
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112308853A (en) * | 2020-10-20 | 2021-02-02 | 平安科技(深圳)有限公司 | Electronic equipment, medical image index generation method and device and storage medium |
WO2021189914A1 (en) * | 2020-10-20 | 2021-09-30 | 平安科技(深圳)有限公司 | Electronic device, medical image index generation method and apparatus, and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP2006500124A (en) | Method and system for reading medical images with a computer aided detection (CAD) guide | |
CN111062947A (en) | Deep learning-based X-ray chest radiography focus positioning method and system | |
US10832410B2 (en) | Computer system, method, and program for diagnosing subject | |
CN109003269A (en) | A kind of mark extracting method for the medical image lesion that can improve doctor's efficiency | |
CN112862778A (en) | Scanning image detection method, electronic device and computer readable medium | |
WO2021073117A1 (en) | Oct image segmentation method, apparatus and device, and storage medium | |
CN111382801B (en) | Medical image classification method, device, equipment and storage medium | |
CN111508016A (en) | Vitiligo regional chromatic value and area calculation method based on image processing | |
CN112579808A (en) | Data annotation processing method, device and system | |
Ghorakavi | TBNet: pulmonary tuberculosis diagnosing system using deep neural networks | |
KR20180045473A (en) | System, method and computer program for melanoma detection using image analysis | |
KR102095730B1 (en) | Method for detecting lesion of large intestine disease based on deep learning | |
CN111724347A (en) | Chest radiography image anomaly detection method and system based on deep learning | |
Lucassen et al. | Deep learning for detection and localization of B-lines in lung ultrasound | |
CN114757942A (en) | Method for detecting active tuberculosis by multilayer spiral CT (computed tomography) based on deep learning | |
CN111401102A (en) | Deep learning model training method and device, electronic equipment and storage medium | |
CN110197722B (en) | AI-CPU system platform | |
CN113537407B (en) | Image data evaluation processing method and device based on machine learning | |
Ramaraj et al. | Histogram variance controlled bleeding detectors for wireless capsule endoscopic images | |
Ashame et al. | Abnormality Detection in Eye Fundus Retina | |
CN113222984A (en) | X-ray film new coronary pneumonia identification and detection method based on fine-grained image identification and target detection algorithm | |
KR20190113351A (en) | Analysis apparatus for medical image using machine learning algorithm | |
CN111612755B (en) | Lung focus analysis method, device, electronic equipment and storage medium | |
Dilna et al. | A novel method for bleeding detection in Wireless Capsule Endoscopic images | |
Tang | Discovering Unknown Diseases with Explainable Automated Medical Imaging |
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 | ||
WW01 | Invention patent application withdrawn after publication | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20200929 |