CN115272328B - Lung ultrasonic image detection model training system for new coronary pneumonia - Google Patents
Lung ultrasonic image detection model training system for new coronary pneumonia Download PDFInfo
- Publication number
- CN115272328B CN115272328B CN202211187174.4A CN202211187174A CN115272328B CN 115272328 B CN115272328 B CN 115272328B CN 202211187174 A CN202211187174 A CN 202211187174A CN 115272328 B CN115272328 B CN 115272328B
- Authority
- CN
- China
- Prior art keywords
- model
- image
- training
- lung
- symptom
- 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
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- 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
- G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing
-
- 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
- G06T2207/10132—Ultrasound 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
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Radiology & Medical Imaging (AREA)
- General Health & Medical Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Epidemiology (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Ultra Sonic Daignosis Equipment (AREA)
Abstract
The invention provides a lung ultrasonic image detection model training system for new coronary pneumonia, which comprises: the image processing module is used for respectively processing the acquired original lung ultrasonic image sets according to at least one image processing method to obtain at least one preprocessing image set corresponding to the at least one image processing method; the image labeling module is used for labeling each preprocessed image set by at least one labeling method to obtain at least one sample set; and the model training module is used for training at least one model to be trained through the sample set for each sample set to obtain at least one trained model, evaluating the at least one trained model according to a preset evaluation standard, determining at least one target model and determining an image processing method of the sample set corresponding to the target model. The direct detection of the focus is realized, no radioactive ray damage is caused, and the method has obvious advantages compared with detection technologies such as CT and the like, and has higher application value and popularization value.
Description
Technical Field
The invention relates to the technical field of lung ultrasonic image symptom detection, in particular to a lung ultrasonic image detection model training system for new coronary pneumonia.
Background
The novel coronavirus pneumonia (new coronary pneumonia for short, COVID-19 for short) has strong infectivity and rapid disease progression. The existing new coronary pneumonia detection depends on nucleic acid and antibody detection reagents, the detection period is long, and the lung symptoms cannot be directly diagnosed. And the efficiency of doctor diagnosis and treatment of patients and the survival rate of patients are improved by accurately and efficiently detecting the lung ultrasonic image symptoms. Therefore, a simple and practical model training device or system for early screening and dynamic monitoring is urgently needed clinically.
Disclosure of Invention
The invention provides a lung ultrasonic image detection model training system for new coronary pneumonia, which is used for solving the defects that the detection period is long and the lung symptoms cannot be directly diagnosed in the prior art and realizing early screening of the lung symptoms by a simple, convenient and practical device or system.
The invention provides a lung ultrasonic image detection model training system for new coronary pneumonia, which comprises: the image processing module is used for respectively processing the acquired original lung ultrasonic image sets according to at least one image processing method to obtain at least one preprocessing image set corresponding to the at least one image processing method; the image labeling module is used for labeling each preprocessed image set by at least one labeling method to obtain at least one sample set; the model training module is used for training at least one model to be trained through the sample set for each sample set to obtain at least one trained model, evaluating the at least one trained model according to a preset evaluation standard, determining at least one target model and determining an image processing method of the sample set corresponding to the target model, wherein the at least one target model corresponds to at least one pulmonary symptom, and the image processing method of the sample set corresponding to the target model is used for processing the ultrasonic image to be detected.
Further, the image processing method includes at least one of edge detection, inverse graying, binarization, and normalization processing.
Further, the labeling method comprises at least one of loose labeling and compact labeling, wherein the loose labeling refers to a region where the bounding box in the preprocessed image contains the image region of the skin edge contacted by the ultrasonic probe and the lung symptom, and the compact labeling refers to a region where the bounding box in the preprocessed image contains the lung symptom.
The system further comprises an image preprocessing module, an image processing method and a preprocessing module, wherein the image preprocessing module is used for acquiring the ultrasonic image to be detected and at least one target model, acquiring a sample set corresponding to the target model, and processing the ultrasonic image to be detected through the image processing method to obtain a preprocessed image.
The system further comprises a detection module, which is used for inputting the preprocessed image into the at least one target model, and obtaining at least one detection result according to a preset confidence degree, wherein the at least one detection result corresponds to at least one lung symptom.
Further, the detection module is further configured to input the preprocessed image into at least one target model to obtain at least one first detection result, where the first detection result corresponds to at least one pulmonary symptom; and weighting and summing the at least one first detection result according to the lung symptoms to obtain at least one final detection result corresponding to the at least one lung symptom.
Further, the first detection result comprises at least one symptom, each symptom is represented as a quintuple, and the quintuple comprises an image name, a lung symptom, a serial number, coordinates of the upper left corner and the lower right corner of a symptom enclosure box and a confidence level.
Further, the ultrasonic diagnosis device also comprises a counting module used for counting the number of at least one lung symptom existing in the ultrasonic image to be detected according to the at least one detection result.
Further, the statistical module is further configured to perform alignment processing on an area where the symptom enclosure frame is located; calculating the intersection ratio between every two areas, merging the areas with the intersection ratio larger than a specified threshold value into a symptom as a repeated picture, recording the areas with the intersection ratio smaller than the specified threshold value as two symptoms, and finishing deduplication processing; and counting the number of at least one lung symptom in the ultrasonic image to be detected.
The invention provides a lung ultrasonic image detection model training system for new coronary pneumonia, which is used for respectively processing an obtained original lung ultrasonic image set according to at least one image processing method through an image processing module to obtain at least one preprocessed image set corresponding to the at least one image processing method; the image labeling module is used for labeling each preprocessed image set by at least one labeling method to obtain at least one sample set; the model training module is used for training at least one model to be trained through the sample set for each sample set to obtain at least one trained model, evaluating the at least one trained model according to a preset evaluation standard, determining at least one target model and determining an image processing method of the sample set corresponding to the target model, wherein the at least one target model corresponds to at least one pulmonary symptom, and the image processing method of the sample set corresponding to the target model is used for processing the ultrasonic image to be detected. The method realizes the combination generation of a plurality of sample sets by adopting different preprocessing modes and different labeling methods and trains a plurality of versions of models to perform model optimization.
Drawings
In order to more clearly illustrate the present invention or the technical solutions in the prior art, the drawings used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a schematic diagram of some embodiments of a pulmonary ultrasound image detection model training system for new coronary pneumonia provided in accordance with the present invention;
fig. 2 is a schematic diagram of the image after Canny, inverse graying, binarization and normalization preprocessing.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1, fig. 1 is a schematic diagram of a lung ultrasound image detection model training system for new coronary pneumonia according to some embodiments of the present invention. As shown in fig. 1, the system includes the following modules:
the image processing module 100 is configured to process the acquired original lung ultrasound image sets according to at least one image processing method, respectively, to obtain at least one preprocessed image set corresponding to the at least one image processing method.
In some optional implementations, the image processing method includes at least one of edge detection, inverse graying, binarization, and normalization processes.
And an image labeling module 200, configured to label each preprocessed image set by at least one labeling method, respectively, to obtain at least one sample set.
In some alternative implementations, the labeling method includes at least one of a loose type labeling and a compact type labeling, wherein the loose type labeling refers to a region where a bounding box in the preprocessed image contains an image region of a skin edge contacted by the ultrasound probe and a region where the lung symptom is located, and the compact type labeling refers to a region where the bounding box in the preprocessed image contains the lung symptom.
As an example, lung ultrasound images are image processed to generate a plurality of sample sets, including:
A1. acquiring an original picture set, and respectively carrying out Canny, inverse graying, binarization and normalization pretreatment on the picture to obtain five picture sets including an original picture;
A2. respectively carrying out loose type labeling and compact type labeling on the five picture sets to obtain ten sample sets which are marked as W1-W10;
A3. and dividing the ten sample sets into a training set and a verification set according to a specified proportion, wherein the training set is used for model training, and the verification set is used for model evaluation.
The model training module 300 is configured to, for each sample set, train at least one model to be trained through the sample set to obtain at least one trained model, evaluate the at least one trained model according to a preset evaluation criterion, determine at least one target model, and determine an image processing method of the sample set corresponding to the target model, where the at least one target model corresponds to at least one pulmonary symptom, and the image processing method of the sample set corresponding to the target model is used to process an ultrasound image to be detected.
In some embodiments, the original target detection model (i.e., the model to be trained) is trained using different sample sets, which includes the following steps:
B1. configuring pre-trained initial weight and hyper-parameters;
B2. training a target detection model Ui,1= < i < =10 with each sample set Wi; the method comprises the following specific steps:
b2.1 training the model Ui by using a Wi training set to obtain model loss lossi;
b2.2, judging whether the loss lossi is converged to reach a preset standard or not, if not, adjusting the hyper-parameters of the model, and retraining the model; if the preset standard is reached, finishing the training and deriving a trained model weight coefficient;
b2.3, inputting the verification set of Wi into the trained model, and calculating the evaluation indexes of the model through the prediction result and the labeling result, wherein the evaluation indexes comprise the detection accuracy Pi and the recall rate Ri;
b2.4, if the evaluation index of the model does not reach the preset standard, adjusting the hyper-parameter of the model, and retraining the model; and if the preset standard is reached, finishing the training and deriving the trained model weight coefficient.
In some embodiments, the trained target detection models U1 to U10 (i.e., at least one target model) are optimized, and the specific steps are as follows:
C1. aiming at four symptoms of an A line, a B line, lung consolidation change C and lung hydrops PE, comparing evaluation indexes of ten versions of models, selecting a model with the optimal index corresponding to each symptom as an optimal model, and marking as A _ U, B _ U, C _ U and PE _ U respectively;
C2. the sample sets used for training the models A _ U, B _ U, C _ U, PE _ U are denoted as DA, DB, DC, DPE.
In some optional implementation manners, the new coronary pneumonia-oriented lung ultrasound image detection model training system further includes an image preprocessing module, configured to acquire an ultrasound image to be detected and at least one target model, and an image processing method for acquiring a sample set corresponding to the target model, and the image processing method is used to process the ultrasound image to be detected to obtain a preprocessed image.
The lung ultrasonic image detection model training system for the new coronary pneumonia further comprises a detection module, wherein the detection module is used for inputting the preprocessed image into the at least one target model and obtaining at least one detection result according to a preset confidence coefficient, and the at least one detection result corresponds to at least one lung symptom.
The detection module is further configured to input the preprocessed image into at least one target model to obtain at least one first detection result, where the first detection result corresponds to at least one pulmonary symptom; and weighting and summing the at least one first detection result according to the pulmonary symptoms to obtain at least one final detection result corresponding to the at least one pulmonary symptom.
The first detection result comprises at least one symptom, each symptom is represented as a quintuple, and the quintuple comprises an image name, lung symptoms, a serial number, coordinates of the upper left corner and the lower right corner of a symptom enclosure frame and confidence.
The lung ultrasonic image detection model training system for the new coronary pneumonia further comprises a counting module, wherein the counting module is used for counting the number of at least one lung symptom existing in the ultrasonic image to be detected according to the at least one detection result.
In some embodiments, obtaining an ultrasound image of a lung of a patient, and performing inference on all the images to obtain a detection result includes the following specific steps:
D1. acquiring an ultrasound image of the lung of a detected person, numbering according to a preset region dividing mode, and naming an image file by region numbering and sequence number;
D2. reasoning is carried out on each image to obtain a reasoning result, and the method comprises the following specific steps:
d2.1, respectively preprocessing the image according to the preprocessing modes of the sample sets DA, DB, DC and DPE;
d2.2, inputting the processed image into corresponding models A _ U, B _ U, C _ U and PE _ U to obtain an inference result;
d2.3, taking the result with the maximum confidence coefficient as the reasoning result of the image;
d2.4 the reasoning result of the image contains 0 or more symptoms, each symptom is expressed as a quintuple and comprises an image name, the symptom, a serial number, coordinates of the upper left corner and the lower right corner of a symptom enclosing frame and confidence;
D3. respectively counting the number of B lines and the number of lung consolidation inferred from the lung image of each side of the patient, and specifically comprising the following steps:
d3.1, respectively counting the number of B lines and the number of lung real changes inferred from the lung image of each region of the patient;
and D3.2, respectively adding the number of the B lines and the number of the lung capacity changes in each region of each side lung image to obtain the number of the B lines and the number of the lung capacity changes of the side lung image.
In some optional implementation manners, the statistical module is further configured to perform alignment processing on an area where the symptom enclosure box is located; calculating the intersection ratio between every two areas, combining the areas with the intersection ratio larger than a specified threshold value as a repeated picture into a symptom, recording the areas with the intersection ratio smaller than the specified threshold value as two symptoms, and finishing duplicate removal processing; and counting the number of at least one lung symptom in the ultrasonic image to be detected.
In some embodiments, the number of B lines and the number of lung real changes inferred from the lung image of each region of the patient are counted, which is specifically implemented as follows:
d3.1.1, aligning all pictures in the region;
calculating the intersection ratio between every two D3.1.2 pictures, merging the pictures with the intersection ratio larger than a specified threshold value into a symptom as a repeated picture, recording the pictures with the intersection ratio smaller than the specified threshold value as two symptoms, and finishing the duplicate removal treatment;
and D3.1.3, counting the number of B lines and lung capacity change in the region.
The invention aims to realize a lung ultrasonic image detection model training system for new coronary pneumonia, which can realize the following steps: 1) Carrying out image processing on the lung ultrasonic image to generate a plurality of sample sets; 2) Training a target detection model by using different sample sets; 3) Selecting the target detection models of different versions; 4) Acquiring an ultrasonic image of the lung of a patient, and reasoning all the images to obtain a detection result; 5) And performing risk assessment by using the number of the B lines and the lung consolidation number of the two side lungs of the patient, epidemiological information and the patient self-describing symptom information. The invention automatically completes the detection and judgment of the pneumonia ultrasonic image by using the related technology of computer image processing, and completes the storage and output of the detection result. The invention can be used for risk assessment of new coronary pneumonia, directly detects the focus, obtains a detection result in real time, is simple, convenient and easy to learn, has no radioactive ray damage, has obvious advantages compared with detection technologies such as CT and the like, and has higher application value and popularization value.
The present invention is further described below by referring to data examples according to the following steps, taking a set of experimental data as an example (268 pieces of original images of lung ultrasound images are used, and the target detection model used is the YOLOv5 model):
the lung ultrasonic image is subjected to image processing to generate a plurality of sample sets, and the method specifically comprises the following steps:
1.1, acquiring an original image set, respectively carrying out Canny, inverse graying, binarization and normalization pretreatment on the image, and cutting after the pretreatment, as shown in FIG. 2;
1.2, performing loose type labeling and compact type labeling on the five image sets respectively to obtain ten sample sets in total; the loose type marking finger bounding box comprises an image area of the skin edge contacted by the ultrasonic probe and an area where the symptom is located, and the compact type marking finger bounding box only comprises the area where the symptom is located;
1.3 ten sample sets were all scaled to 9:1, dividing the training set into a training set and a verification set, wherein the training set is used for model training, and the verification set is used for model evaluation;
1.4 training a target detection model by using different sample sets;
1.5, selecting different versions of target detection models optimally, comparing evaluation indexes of ten versions of models aiming at four symptoms of an A line, a B line, lung consolidation C (or called C consolidation symptom) and lung consolidation PE (or called PE consolidation symptom), selecting a model with the highest accuracy corresponding to each symptom as an optimal model of the symptom, and selecting a preprocessing and labeling mode of a sample set corresponding to the optimal model as an optimal preprocessing and labeling mode; the example results are shown in table 1, the optimal model of the effusion symptoms of line a and PE is U1, the corresponding image preprocessing mode is inverse gray scale, the labeling mode is compact labeling, the optimal model of the real variable symptoms of line B and C is U2, the corresponding image uses original images, and the labeling mode is compact labeling;
TABLE 1 model parameters and accuracy
1.6 acquiring an ultrasonic image of the lung of the patient, and reasoning all the images to obtain that the left lung of the patient contains 1 lung consolidation change and 0B line, and the right lung contains 2 lung consolidation changes and 1B line;
as an example, the risk assessment can be performed by using the number of B lines and lung consolidation number on both sides of the patient, epidemiological information and patient self-describing symptom information.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. Based on the understanding, the above technical solutions substantially or otherwise contributing to the prior art may be embodied in the form of a software product, which may be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the various embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (9)
1. The utility model provides a lung ultrasonic image detection model training system towards new coronary pneumonia which characterized in that includes:
an image processing module, configured to process the acquired original lung ultrasound image sets according to at least one image processing method, respectively, to obtain at least one preprocessed image set corresponding to the at least one image processing method, including: acquiring an original lung ultrasonic image set, and respectively carrying out Canny, inverse graying, binarization and normalization pretreatment on the original lung ultrasonic image to obtain a plurality of pretreatment image sets;
the image labeling module is used for labeling each preprocessed image set by at least one labeling method to obtain at least one sample set, and comprises:
respectively carrying out loose type labeling and compact type labeling on the multiple preprocessed image sets to obtain multiple sample sets; the model training module is used for training at least one model to be trained through the sample set for each sample set to obtain at least one trained model, evaluating the at least one trained model according to a preset evaluation standard, determining at least one target model and determining an image processing method of the sample set corresponding to the target model, wherein the at least one target model corresponds to at least one pulmonary symptom, and the image processing method of the sample set corresponding to the target model is used for processing an ultrasonic image to be detected;
training a model to be trained by using different sample sets, and specifically comprising the following steps of:
B1. configuring pre-trained initial weight and hyper-parameters;
B2. training a target detection model Ui,1= < i < =10, with each sample set Wi; the method comprises the following specific steps:
b2.1 training the model Ui by using a Wi training set to obtain model loss lossi;
b2.2, judging whether the loss lossi is converged to reach a preset standard, if not, adjusting the hyper-parameter of the model, and retraining the model; if the preset standard is reached, finishing the training and deriving a trained model weight coefficient;
b2.3, inputting the verification set of Wi into the trained model, and calculating the evaluation indexes of the model through the prediction result and the labeling result, wherein the evaluation indexes comprise the detection accuracy Pi and the recall rate Ri;
b2.4, if the evaluation index of the model does not reach the preset standard, adjusting the hyper-parameter of the model, and retraining the model; and if the preset standard is reached, finishing the training and deriving the trained model weight coefficient.
2. The system for training the pulmonary ultrasound image detection model for new coronary pneumonia according to claim 1, wherein the image processing method comprises at least one of edge detection, inverse graying, binarization and normalization processing.
3. The system for training the pulmonary ultrasound image detection model for new coronary pneumonia according to claim 1, wherein the labeling method comprises at least one of loose type labeling and compact type labeling, wherein the loose type labeling means that the bounding box in the preprocessed image contains the image region of the skin edge contacted by the ultrasound probe and the region where the pulmonary symptom is located, and the compact type labeling means that the bounding box in the preprocessed image contains the region where the pulmonary symptom is located.
4. The system for training the new coronary pneumonia-oriented pulmonary ultrasound image detection model according to any one of claims 1-3, further comprising an image preprocessing module for acquiring an ultrasound image to be detected and at least one target model, and an image processing method for acquiring a sample set corresponding to the target model, wherein the image processing method is used for processing the ultrasound image to be detected to obtain a preprocessed image.
5. The system for training the lung ultrasound image detection model for new coronary pneumonia according to claim 4, further comprising a detection module for inputting the preprocessed image into the at least one target model to obtain at least one detection result according to a preset confidence level, wherein the at least one detection result corresponds to at least one lung symptom.
6. The system for training the lung ultrasound image detection model for new coronary pneumonia according to claim 4, wherein the detection module is further configured to input the preprocessed image into at least one target model to obtain at least one first detection result, and the first detection result corresponds to at least one lung symptom; and weighting and summing the at least one first detection result according to the pulmonary symptoms to obtain at least one final detection result corresponding to the at least one pulmonary symptom.
7. The system for training the pulmonary ultrasound image detection model for new coronary pneumonia according to claim 6, wherein the first detection result comprises at least one symptom, each symptom is represented by a quintuple comprising an image name, a pulmonary symptom, a serial number, coordinates of upper left corner and lower right corner of a symptom bounding box, and a confidence level.
8. The system for training the ultrasonic image detection model of lung for new coronary pneumonia according to claim 5 or 6, characterized by further comprising a statistic module for counting the number of at least one lung symptom existing in the ultrasonic image to be detected according to the at least one detection result.
9. The system for training the lung ultrasonic image detection model for new coronary pneumonia according to claim 8, wherein the statistical module is further configured to align the region where the symptom enclosure box is located; calculating the intersection ratio between every two areas, combining the areas with the intersection ratio larger than a specified threshold value as a repeated picture into a symptom, recording the areas with the intersection ratio smaller than the specified threshold value as two symptoms, and finishing duplicate removal processing; and counting the number of at least one lung symptom in the ultrasonic image to be detected.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211187174.4A CN115272328B (en) | 2022-09-28 | 2022-09-28 | Lung ultrasonic image detection model training system for new coronary pneumonia |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211187174.4A CN115272328B (en) | 2022-09-28 | 2022-09-28 | Lung ultrasonic image detection model training system for new coronary pneumonia |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115272328A CN115272328A (en) | 2022-11-01 |
| CN115272328B true CN115272328B (en) | 2023-01-24 |
Family
ID=83756747
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211187174.4A Active CN115272328B (en) | 2022-09-28 | 2022-09-28 | Lung ultrasonic image detection model training system for new coronary pneumonia |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115272328B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2005098739A1 (en) * | 2004-04-08 | 2005-10-20 | Mobileye Technologies Limited | Pedestrian detection |
| CN109461495A (en) * | 2018-11-01 | 2019-03-12 | 腾讯科技（深圳）有限公司 | A kind of recognition methods of medical image, model training method and server |
| CN111815608A (en) * | 2020-07-13 | 2020-10-23 | 北京小白世纪网络科技有限公司 | New coronary pneumonia patient recovery time prediction method and system based on deep learning |
| CN112132800A (en) * | 2020-09-18 | 2020-12-25 | 上海市肺科医院 | Pulmonary fibrosis detection and severity evaluation method and system based on deep learning |
-
2022
- 2022-09-28 CN CN202211187174.4A patent/CN115272328B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2005098739A1 (en) * | 2004-04-08 | 2005-10-20 | Mobileye Technologies Limited | Pedestrian detection |
| CN109461495A (en) * | 2018-11-01 | 2019-03-12 | 腾讯科技（深圳）有限公司 | A kind of recognition methods of medical image, model training method and server |
| CN111815608A (en) * | 2020-07-13 | 2020-10-23 | 北京小白世纪网络科技有限公司 | New coronary pneumonia patient recovery time prediction method and system based on deep learning |
| CN112132800A (en) * | 2020-09-18 | 2020-12-25 | 上海市肺科医院 | Pulmonary fibrosis detection and severity evaluation method and system based on deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115272328A (en) | 2022-11-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kollias et al. | Ai-mia: Covid-19 detection and severity analysis through medical imaging | |
| CN108464840B (en) | Automatic detection method and system for breast lumps | |
| US5873824A (en) | Apparatus and method for computerized analysis of interstitial infiltrates in chest images using artificial neural networks | |
| Islam et al. | Automatic detection of pneumonia on compressed sensing images using deep learning | |
| CN112150442A (en) | New crown diagnosis system based on deep convolutional neural network and multi-instance learning | |
| Yao et al. | CT grading of lung disease in lymphangioleiomyomatosis | |
| Koprowski et al. | Assessment of significance of features acquired from thyroid ultrasonograms in Hashimoto's disease | |
| Dovganich et al. | Automatic out-of-distribution detection methods for improving the deep learning classification of pulmonary X-ray images | |
| Gundreddy et al. | Assessment of performance and reproducibility of applying a content‐based image retrieval scheme for classification of breast lesions | |
| Cenggoro et al. | Systematic literature review: An intelligent pulmonary TB detection from chest x-rays | |
| CN115272328B (en) | Lung ultrasonic image detection model training system for new coronary pneumonia | |
| CN117727441A (en) | Method for predicting lung cancer immune curative effect based on clinical-fusion image computer model | |
| JP2023143875A (en) | Medical image analysis method | |
| Chen et al. | Classification of glomerular spikes using Convolutional Neural Network | |
| Wan et al. | Depth-wise convolution with attention neural network (DWA) for pneumonia detection | |
| Morani | COVID-19 detection using segmentation, region extraction and classification pipeline | |
| CN113450306B (en) | Method for providing fracture detection tool | |
| CN114998203A (en) | System and method for accurately diagnosing occupational pneumoconiosis based on artificial intelligence | |
| Suganyadevi | DarkCVNet: Optimized Pneumonia and Covid 19 detection using CXR Images | |
| Dhanapala et al. | Transfer learning techniques with SVM for covid-19 disease prediction based on chest X-ray images | |
| CN111325282A (en) | Mammary gland X-ray image identification method and device suitable for multiple models | |
| Chu | Machine learning for automation of Chromosome based Genetic Diagnostics | |
| Ye et al. | A deep convolutional neural network based hybrid framework for fetal head standard plane identification | |
| Joharah et al. | Evaluation of fetal head circumference (HC) and biparietal diameter (BPD (biparietal diameter)) in ultrasound images using multi-task deep convolutional neural network | |
| CN112766333B (en) | Medical image processing model training method, medical image processing method and device |
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 |